cyclic definition with 'ans'

I'm not at home before wednesday so I can't see the problem (no linux box here), and there is something I don't understand.

You say "variables should not be replaced UNLESS it is the 'ans' variable" but I'm not agree.

If ans=15791, 170.07*ln[ans] should return at less (since we are here in simplification mode) 170.07*ln[15791]
and so, no recursive assignment !! And more, since a part of the expression is a floating point number, we can evaluate the all expression.

Could you tell me what's happening with the following example ?

::> z=4451+11340
::> z=170.07*ln[z]
::> z
Do you have the same "oops", an error with a "recursive assignment" or the good result ?

Thanks.

Hi, David.

::> z=4451+11340
ans = 15791
::> z=170.07*ln[z]
IDL:Numexp/EvalError:1.0: recursive assignment

And, yes, I agree that we can evaluate 170.07*ln[anything] as 170.07 is
already a rational number.

CHeers
Alberto

David Boucher wrote:
> I'm not at home before wednesday so I can't see the problem (no linux
> box here), and there is something I don't understand.
>
> You say "variables should not be replaced UNLESS it is the 'ans'
> variable" but I'm not agree.
>
> If ans=15791, 170.07*ln[ans] should return at less (since we are here in simplification mode) 170.07*ln[15791]
> and so, no recursive assignment !! And more, since a part of the expression is a floating point number, we can evaluate the all expression.
>
> Could you tell me what's happening with the following example ?
>
> ::> z=4451+11340
> ::> z=170.07*ln[z]
> ::> z
> Do you have the same "oops", an error with a "recursive assignment" or the good result ?
>
> Thanks.
>
> ** Changed in: numexp-core (upstream)
> Importance: Undecided => High
> Assignee: (unassigned) => David Boucher
>

--
Alberto Simões - Departamento de Informática - Universidade do Minho
                  Campus de Gualtar - 4710-057 Braga - Portugal

   If you have the technical capability to complain about the docs,
   you have the technical ability to improve those docs.