Are you sure your file is a comma delimited file like you told PROC IMPORT it was? Take a look at the file with a text editor. Or just use a simple data step to look at the file.  For example this data step will read the first two lines and dump them to the LOG.  If any of the characters on the lines are un-printable then it will also show the hexadecimal code for all of the characters in that part of the line. data _null_; infile "D:\Rakee\E-UL-211606-DataRequest_2013.csv" obs=2; input; list; run;   ... View more

How is that the case?  Unlike the other "cat" functions, I understand that CAT does not strip blanks - i.e. it acts like the "||" operator, so I do not expect it to create collisions prior to the MD5 function.   And that's why I didn't bother with creating arbitrary separators between the concatenated fields. True, my bad. But then you potentially have a very long string to hash. It's still better to trim and separate imho.   Agreed.  Although I hadn't considered _MD5 ($16) long.  But see comment to #4:  The issue we're trying to avoid here is running out of memory (at the cost of CPU, used to derive the hash). Having said that, SAS hash tables are quite wasteful, and I just checked that a KEY length of 16 and DATA length of 16 occupies 64 bytes per hash table item, just as long as a KEY length of 16 and DATA length of 1. So that's one of my "improvements" gone.   ... View more