WHITLOI1 <WHITLOI1@WESTAT.COM> wrote:

> Alan Rimm-Kaufman <rimmkaufman@CRUTCHFIELD.COM> asked about the > NODUPKEY option of PROC SORT and suggested that it might not work. The > The NODUPKEY works as advertised. It makes the the BY-GROUP values > unique by removing all records after the first with the same key > values. Since the keys are unique, the records are also unique.

I haven't encountered problems with it myself, but one of the Usage Notes I posted earlier says that NODUPKEY has problems under some circumstances. I think it's a good idea to avoid both options, and do the duplicate deletion in a data step or with PROC SQL (which may force you to rethink what you want to do, sometimes a good thing).

> On the other hand, the NODUP option is a long time snare! I first > remember it in version 5.03. And there is no relief in sight, since > it remains in the version 7 beta.

That's not good. SAS specifically said that some of the SORT problems would be fixed in V7. Perhaps they were referring to other sort problems.