Re: ROUTINE_NAMES and other magic

[Craig Markwardt <craigmnet(at)cow.physics.wisc.edu>]

"R.Bauer" <R.Bauer@fz-juelich.de> writes:
> Craig Markwardt wrote:
> 
> > One thing that ROUTINE_NAMES() (**note) cannot do is *add* variables
> > to another level.  If the variable exists, then you can muck as much
> > as you want with it, but if it doesn't exist, sorry.
> 
> You can add variables to another level.
> Try this!
> 
> 
> 
> PRO DEEPSTOP, level
>    IF level EQ 1 THEN BEGIN
>       level = ROUTINE_NAMES(/LEVEL)
>       varName = 'A'
>       void = ROUTINE_NAMES(varName, STORE=(level+1), 8)
>    ENDIF
>    IF level EQ 2 THEN begin
>    print,a
>    STOP
>    end
>    deepstop, level + 1
> END

Ahhh, but I argue that your procedure works for the reasons I said
before.  The variable A already existed in the procedure because you
used it in a statement ("print, a").  So you really were not *adding*
the variable to the procedure.

Try this one:

PRO DEEPSTOP2, level
   IF level EQ 1 THEN BEGIN
;     a = 0
      level = ROUTINE_NAMES(/LEVEL)
      varName = 'A'
      void = ROUTINE_NAMES(varName, STORE=(level+1), 8)
      help
   ENDIF
   IF level EQ 2 THEN begin
     help
     STOP
   end
   deepstop, level + 1
END

This procedure does not mention "A" explicitly anywhere, and I can't
get beyond the first IF clause.

IDL> deepstop2, 1
% ROUTINE_NAMES: Variable is undefined: A.
% Execution halted at:  DEEPSTOP2           5 /dev/tty
%                       $MAIN$                 

But, if you uncomment the "a=0" line above, then you can get further.
What I find is that the value of A is set at *both* levels!

I am using an older version of IDL, 5.2.  This tells me that the
functionality of ROUTINE_NAMES continued to evolve between versions,
and that you can't be guaranteed to be able to add new variables in
older versions.

IDL> help, !version, /str
** Structure !VERSION, 5 tags, length=80:
   ARCH            STRING    'alpha'
   OS              STRING    'OSF'
   OS_FAMILY       STRING    'unix'
   RELEASE         STRING    '5.2'
   BUILD_DATE      STRING    'Oct 30 1998'

Craig


-- 
--------------------------------------------------------------------------
Craig B. Markwardt, Ph.D.         EMAIL:    craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
--------------------------------------------------------------------------