Thanks, Mike. I remembered hearing about the concept, but I could not remember that it was implemented as a function, let alone the name of the function. So I just used arbitrary but presumably safe constants.

So make it

if losize=0 then losize = -constant('small'); if hisize=. then hisize = constant('big');

Note the more important correction to my erlier code. The zero is replaced with a tiny *negative* value.

On Tue, 14 Jan 2003 17:37:40 -0500, Mike Rhoads <RHOADSM1@WESTAT.COM> wrote:

>Howard Schreier wrote (in very small part): >... > if losize=0 then losize = 1e-9; > if hisize=. then hisize = 1e50; >... ><end quote> > >Not that it will make any practical difference to the solution of this >problem, but it does provide a chance to point out the relatively new >CONSTANT function (he said oxymoronically), which returns a variety of >useful mathematical or machine-specific numeric constant values, depending >on the character string passed to it. Of particular importance here are >CONSTANT('BIG'), which "returns the largest double-precision floating point >number (8-bytes) that is representable on your computer," and >CONSTANT('SMALL'), which "returns the smallest double-precision floating >point number (8-bytes) that is representable on your computer." > >More info is available from the SAS V8 Language Reference Dictionary, which >is the source of the quotes above. > >Mike Rhoads >Westat >RhoadsM1@Westat.com