Date: Thu, 4 Jan 2007 14:59:37 -0500
Reply-To: Dianne Rhodes <diannerhodes@COMCAST.NET>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Dianne Rhodes <diannerhodes@COMCAST.NET>
Subject: Re: Is "NOT IN" an undocumented operator?
On Thu, 4 Jan 2007 14:20:07 -0500, Mike Rhoads <RHOADSM1@WESTAT.COM> wrote:
>I was modifying a program written by somebody else earlier today, and
>came across an IF-statement using "NOT IN". I thought I had remembered
>at one point that this did not work, and so decided to (1) verify
>whether it worked, and (2) check the documentation. The code below does
>in fact work (in SAS 9.1.3 under WinXP):
>LENGTH Char1 $ 1 Result $ 5;
>DO N = RANK('A') TO RANK('Z');
> Char1 = BYTE(N);
> IF Char1 NOT IN ('A','E','I','O','U') THEN Result = 'False';
> ELSE Result = 'True';
> PUT Char1= Result=;
>According to the documentation, unless I'm either missing or misreading
>something, I am surprised that this works. "IN" is documented as an
>infix operator requiring an expression on the left and a list of values
>on the right (and "Char1 NOT" is not an expression). "NOT" is
>documented as a prefix operator that applies to the logical expression
>that follows (and IN... by itself is not a logical expression). I
>thought I had remembered that you had to write this like so:
>IF NOT(Char1 IN ('A','E','I','O','U')) ...
>which is more awkward but does match the documented behavior.
>This of course is in contrast to not-equals, which is documented as a
>Not complaining, just wondering.
I have used IF NOT IN(...) for some time. But IN sometimes does some
interpreting on its own, and does not complain. When I was still at
Westat, we came across some code as
IF IN(1-5, 13-17); SAS did not complain. (Would have been version 6.12 or
What it did do is:
IN saw it as IF IN(1, -5, 13, -17) which of course was not what we
wanted. How long that code had been around I don't know. I suspect
someone cut and pasted it from a format value statement, where it would
have worked just fine.
Dianne at BLS