| Re: Going outside the interger limits [message #40953] |
Tue, 21 September 2004 06:26  |
K. Bowman
Messages: 330
Registered: May 2000
|
Senior Member |
|
|
In article <c3f97ff.0409210321.47f8725f@posting.google.com>,
lloyd@evilprofessor.co.uk (Lloyd Watkin) wrote:
> Hi all,
>
> Have been muddling about with this problem for the past weeek, and I'm
> hoping someone can help.
>
> I'm creating an atmospheric transmission model for the sub-mm/far
> infrared region. If I try to calculate the spectrum between 1 and 500
> wavenumbers (numbers are not important) and I set a high resolution,
> the end of the spectrum does not seem to have had spectral line data
> applied to it.
>
> Having looked at where this is happening (at about point number
> 32700), it lies at the point where an integer runs outside it's
> limits. Inside the code, if a spectral line value is negative then it
> gets ignored, hence why I'm assuming (well guessing really) that this
> is the problem.
>
> I have been through my code and converted as much as I can see into
> either a long or a float depending on what it needs to be.
>
> Still getting problems!
>
> I was wondering whether there is anyway to tell in IDL if an integer
> is trying to go outside it's limits? Such as !except = 2, or is there
> a compiler switch which would break the execution if this problem was
> encountered (the latter would be very handy).
>
> Thanks for any help,
>
> Lloyd Watkin
As far as I know, IDL does not report underflow or overflow for integer
data types, and there is no way to force it to do so. You will have to
do it "manually" (e.g., IF (wavenumber LT 0) THEN STOP) and use HELP to
find the culprit.
I put
COMPILE_OPT IDL2
in my startup.pro and in all program units (procedures and functions).
This is equivalent to
COMPILE_OPT DEFINT32, STRICTARR
It changes the default integer type to LONG (32-bit) and requires the
use of square brackets [ ] for array subscripts (to avoid confusion with
function calls).
In my opinion, with modern PCs and workstations, you should have a
really good reason to use INTs instead of LONGs.
Ken Bowman
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
