Date:         Wed, 1 Nov 2000 18:03:54 +0000
Reply-To:     Puddin' Man <pudding_man@POSTMARK.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Puddin' Man <pudding_man@POSTMARK.NET>
Subject:      Re: SAS/Base: automatic variable _n_ behavior
Comments: To: Peter Crawford <peter.crawford@DB.COM>
Content-Type: text/plain; charset="iso-8859-1"

> so what is wrong with a structure I consider to be very clear and why
> say "(the structure should probably be avoided)" ?

Please note that I didn't say "the structure is wrong" ...

> puddin'man say.......
> ""
> It seems, however, I've run across several programs resembling:
>
>   data ...;
>    set ...;
>     ...
>     if <cond1>;
>     ...
>     if <cond2>;
>     ...
>     if  <condition testing _n_> then <take action>;
>     ...
>
> and if folks dealing with such code didn't know when _n_
> incremented, they would likely get into trouble. They may
> anyway (the structure should probably be avoided).
> ""

I dunno that controlling data step iterations by testing
_N_ after subsetting IF's is particularly advisable. If
one is intimate with both the data and the data step
code I suspect there are often better means to the same
intended end. This is, of course, very much a generality.

> I don't expect you have run across any of my code, but the above
> style IMHO wins hands down over the corresponding over-weighty
> nested "if then do end else do end" structures.

I wasn't objecting to subsetting IF's. They are very handy,
very useful.

> It is allways disappointing when testing the end-of-data
> conditions preclude subsetting if statements.

If I understand your statement correctly, I dunno that such
preclusion should ever exist.

Wasn't it Ian that first made a valid point that something like:

data ...;
 set <whatever> end=e1;
 if <condition>;
 if e1 then <action>;

may fail to execute the intended end-of-file action whereas
something like:

 data ...;
 if e1 then <action>;
 set <whatever> end=e1;
 if <condition>;

will not ??

> The prerequisite for encouraging use of the "subsetting if" is
> that maintainers of the code understand the principle.

I agree.

> Surely ( like understanding _n_ ) that is not much to ask ?

Well, it shouldn't be. But you've probably noticed that Q's
re behavior of _N_ have been fairly common on sas-l in
recent years. This is evidence that "not everybody understands
it fully".

> Would you want the applications you create to be "enhanced" by
> maintenance programmers who are learning to program ?

Not if I can hep' it. :-)

  Prosit,
  Puddin'

*******************************************************
*** Puddin' Man     *** Pudding_Man@postmark.net  *****
******************************************************;