I'm out of my depth when it comes to C programming. The following link might give you guidance when it comes to calling a DLL from SAS  SAS(R) 9.3 Companion for Windows You're definitely seeking the challenge here. As much as I know SAS Procedures are using a C compiler. But of course: If you write your code directly in C and for exactly your case then you might very well be able to beat the SAS Proc. If you succeed: Very curious to learn how much you could improve performance. Please add a link (reference) here in this thread once you post your follow-up question/results so that I don't miss it. ... View more

Thanks Patrick. I have the following error message: 9    data want2(drop=_:); 10     set have nobs=nobs; 11     if _n_=1 then 12       do; 13         if 0 then set have have(keep=lat lon rename=(lat=hlat lon=hlon)); 14         dcl hash h1(dataset:'have(keep=recid lat lon rename=(lat=hlat lon=hlon))'); 15         _rc=h1.defineKey('recid'); 16         _rc=h1.defineData('hlat','hlon'); 17         _rc=h1.defineDone(); 18       end; 19     do _i= (recid+1) to nobs; 20       _rc=h1.find(key:_i); 21       geodist=geodist(lat, lon, hlat, hlon, 'm'); 22       output; 23     end; 24   run; NOTE: Variable recid is uninitialized. ERROR: The variable recid in the DROP, KEEP, or RENAME list has never been referenced. ERROR: Hash data set load failed at line 17 column 11. ERROR: DATA STEP Component Object failure.  Aborted during the EXECUTION phase. NOTE: The SAS System stopped processing this step because of errors. NOTE: There were 1 observations read from the data set WORK.HAVE. WARNING: The data set WORK.WANT2 may be incomplete.  When this step was stopped there were 0          observations and 6 variables. NOTE: DATA statement used (Total process time):       real time           0.09 seconds       cpu time            0.01 second EDIT: It worked after I added the "recid" variable: data have; input recid lon lat; cards; 1 -75.547513 39.757077 2 -75.554342 39.749864 3 -75.555394 39.730672 4 -75.556227 39.737546 ;run; Thank you very much! ... View more

Something the Lord made... ... View more

Thank you, all. You guys are amazing! ... View more

Thanks! It works! ... View more

Thanks for getting back to the discussion. I've learned a lot from you! ... View more

Thanks all. The attached "mydata" is the actual data file. Sorry for the confusion. Anyway, by taking Astounding's advice of INPUT statements, I could first read the U1-U4 variables as character, and then convert them into numeric using the INPUT statements: U1new = input(U1, 9.). ... View more

SAnjeev YOu have to learn about the Z. format. The example a few postings earlier in this thread show how much it improves on your solution. ... View more

Thank you. I've learned a lot from you. ... View more

If you only want 8 digit file name ,I prefer to Perl Regular Expression. if prxmatch('/\d{8}/',name) then put name; Ksharp ... View more

Actually, if you do want to preserve the original dataset order then you must do your own remerge as in : proc sql; create table min_x_by_gs as select M.g, x, (x=minX)+(n=1) as min_x_by_g from min_ex as M inner join (select g, min(x) as minX, count(*) as n from min_ex group by g) as S on M.g=S.g; select * from min_x_by_gs; quit; try adding "g1 3" as the last line of min_ex to see the difference. PG ... View more