Date:         Tue, 15 Oct 2002 10:05:08 -0500
Reply-To:     Rodney Sparapani <rsparapa@MCW.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Rodney Sparapani <rsparapa@MCW.EDU>
Organization: Medical College of Wisconsin, Milwaukee
Subject:      Re: sasautos extension macros now tested
Content-Type: text/plain; charset=us-ascii; format=flowed

Roland:

I don't want to be too critical because I know how hard such an endeavor
can be.  In fact,
I applaud your efforts.  But, I find your idea of testing naive.  You've
got some good ideas,
but a couple of days of testing is not really sufficient for a project
like this.  For example,
let's look at %nlobs.  What happens if the data set does not exist?  You
get an ERROR:
message.  That might be a design decision, but I don't think it is a
very good one.  What
would be better is if you returned a value that could be recognized
(personally, I would
use 0, but YMMV).  Also, the user may be interested in counting
observations with
respect to a WHERE clause.  You could add that functionality to %nlobs
or I suppose
you could create %nlobsf.  But, you have to be careful since the nlobsf
parameter wasn't
in 6.12 with the rest of the functionality; it was added at 7.0.  So, in
summary, more
testing, and especially more user testing, is warranted.  If you want to
subscribe to
buzz-words, you could even consider "validation", but I don't go that
far myself.

--
Rodney Sparapani              Medical College of Wisconsin
Sr. Biostatistician           Patient Care & Outcomes Res.
rsparapa@mcw.edu              http://www.mcw.edu/pcor
Was 'Name That Tune' rigged?  WWLD -- What Would Lombardi Do