Arg_Present, XSurface, and Other Assorted Blunders

[davidf(at)dfanning.com (David Fanning)]

Hi Folks,

Just a couple of days ago I was winching and moaning (as
my friends in England say) about Arg_Present and how its
name was just a bit misleading, and how it would lead to
programming errors by people who don't read the documentation,
etc., etc.

Well, don't you know, that dog don't hunt far from home!

I have been getting infrequent, but persistent, reports 
from some of the people who downloaded my new graphics
object program XSURFACE from my web page that it seemed
to hang their IDL session. I could never reproduce the 
problem, so I just chalked it up to the usual whimsical
programming gods having their little joke.

But today I did a clean install of IDL 5.0.2 on my Mac
(I heard it was piling up outside of RSI waiting for
the UPS folks to settle their differences, and sure 
enough it was there waiting for me when I went up to
check out the rumor) and don't you know, the darn
program hung up on me. 

(Well, first it wouldn't run at all, because the default
Mac IDL installation doesn't put *any* IDL directories on
the path. But that is another story, and easily fixed if
you have a firm conviction that every single one of your
programs can't possibly have syntax errors on *every* 
single IDL library function call!)

Anyway, since it hung on the Mac I thought I would try
it on my WindowsNT machine, so I hurried and installed
IDL 5.0.2 there too. When I ran XSURFACE there, it hung too!
But, wait a minute, I hadn't added my program libraries
to the path (the IDL libraries *were* there on my
WindowsNT install, thank goodness). Add my libraries, 
run it again, it works! Hummm...

The only library call I make is to something called LoadData.
I looked at the code near the LoadData call and here is
what I found:

   Catch, error
   IF error NE 0 THEN BEGIN  ; Can't find LoadData.
      data = DIST(41)
      x = Findgen(41)
      y = Findgen(41)
   ENDIF

   IF Arg_Present(data) EQ 0 THEN BEGIN
      data = LoadData(2)
   ENDIF

Here is what this piece of code is *meant* to do:
I want to supply some default data if the user doesn't
pass data into the program in the argument "data".
If I need to create some data I want to use my LoadData
program to get it. But I also know that some people won't
have LoadData, so I want to "Catch" the error that happens
when you try to call a program that is not on your path.
If I catch the error, I create the "data" variable and
continue program execution. 

Because this is a program using IDL 5 specific functionality,
I also wanted to use the new Arg_Present routine to prove
that I am an up-to-date and with-it IDL programmer.
Apparently, I forgot to read the documentation. In any case,
here is what happens.

When the call is first made, "data" is not present and 
Arg_Present reports this correctly. The LoadData error occurs 
and I bounce up to my error handler. I define "data" and 
continue. But Arg_Present *STILL* reports data as missing
in action. This is so even when it is defined AND a variable
that is passed by reference. As a result, my code goes into
an infinite loop.

So now I am confused about exactly what Arg_Present is suppose
to do, but I do know this: it is a grievous mistake to treat
Arg_Present as an function that tells you if an argument 
is present! To fix the problem I swallowed my pride and 
went back to the terribly misnamed N_Elements to solve 
my problem.

If that was *all* I had done, perhaps I could be forgiven.
But in addition I broke one of the cardinal rules of
programming error handlers. I forgot to *always, always*
include some kind of error message in the handler.

If you don't, you will end up blaming RSI for something that
is entirely (well, *almost* entirely) your own fault!

Anyway, some of you want to know how you get to be a renowned
teacher of IDL programming. This is how you do it. Make lots
and lots of foolish programming mistakes. And then when you
get up in front of a class, try hard to remember what it feels 
like to stand there with egg all over your face. :-)

Cheers,

David

----------------------------------------------------------------
David Fanning, Ph.D.              
Fanning Software Consulting
Customizable IDL Programming Courses
Phone: 970-221-0438  E-Mail: davidf@dfanning.com   
Coyote's Guide to IDL Programming: http://www.dfanning.com
IDL 5 Reports: http://www.dfanning.com/documents/anomaly5.html