| Re: idl -quiet command-line option doesn't work in cron [message #67394 is a reply to message #67392] |
Mon, 27 July 2009 14:56   |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
Kenneth P. Bowman wrote:
> In article
> <d012ae88-d64f-4ea7-b015-9efdc089a44c@u16g2000pru.googlegroups.com>,
> Edward Hyer <ejhyer@gmail.com> wrote:
>
>> On Jul 27, 6:26 am, "Kenneth P. Bowman" <k-bow...@null.edu> wrote:
>>> You can turn off floating-point error messages by setting
>>>
>>> !EXCEPT = 0
>>>
>>> but that is not to say it is a good idea. I would think knowing about
>>> floating-point exceptions would be more important than having
>>> "clean" output.
>> Many of my routines ruthlessly and systematically divide by zero,
>> because the denominator is Ndata.
>> Besides, if floating-point errors are so important, how come IDL won't
>> tell me where they happened? :)
>
> As David pointed out, you can set
>
> !EXCEPT = 2
>
> to find out where errors are occurring.
>
> The danger with ignoring floating-point exceptions is that they can
> be occurring in multiple places (some benign, some malignant). With
> the default (!EXCEPT = 1), you only get one notification. This raises
> the possibility of undetected errors.
>
> In my (somewhat obsessive) view, a properly function program should
> not generate floating-point errors, except possibly underflows.
>
> It gives me shivers when someone says, "My program is working fine.
> Just ignore those floating-point errors."
>
> Ken
Of course, graphics can generate them too:
IDL> plot, findgen(11)
% Program caused arithmetic error: Floating illegal operand
Mike
--
www.michaelgalloy.com
Associate Research Scientist
Tech-X Corporation
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
