| Re: BUG in IDL 4.0a PowerMac [message #6038] |
Fri, 05 April 1996 00:00 |
rivers
Messages: 228
Registered: March 1991
|
Senior Member |
|
|
In article <4k3mro$72p@newsreader.wustl.edu>, Paritosh Dhawale <paritosh> writes:
> Hello,
>
> I get this dangerous bug rather randomly running
> IDL 4.0a on a PowerMac 7500. Any one with similar
> problems? IDL fails to return the correct answer
> in statement 4.
>
> IDL> help,yy
> YY FLOAT = 35.0000
> IDL> help,last_row
> LAST_ROW FLOAT = 89.0000
> IDL> print,last_row-yy
> 54.0000
> IDL> print,fix(last_row-yy)
> 53
> IDL> print,fix(last_row)-fix(yy)
> 54
You have really not provided sufficient information. If you created YY and
LAST_ROW by direct assignment, i.e.
YY = 35.0
LAST_ROW = 89.0
then I agree that there is a bug. All machines which I know of represent small
floating point integers exactly, and there should be no roundoff error on
subtraction. However, if YY or LAST_ROW are computed, then they might be
slightly different than the values they appear to have, and you are just seeing
normal roundoff problems. Whenever you are trying to convert a float which
might have some slight roundoff error to an integer like this you should use a
function which returns the nearest integer, not the integer value after
discarding fractional parts.
____________________________________________________________
Mark Rivers (312) 702-2279 (office)
CARS (312) 702-9951 (secretary)
Univ. of Chicago (312) 702-5454 (FAX)
5640 S. Ellis Ave. (708) 922-0499 (home)
Chicago, IL 60637 rivers@cars3.uchicago.edu (Internet)
|
|
|
|
| Re: BUG in IDL 4.0a PowerMac [message #6039 is a reply to message #6038] |
Fri, 05 April 1996 00:00  |
Tim Patterson
Messages: 65
Registered: October 1995
|
Member |
|
|
Paritosh Dhawale wrote:
>
> Hello,
>
> I get this dangerous bug rather randomly running
> IDL 4.0a on a PowerMac 7500. Any one with similar
> problems? IDL fails to return the correct answer
> in statement 4.
>
> IDL> help,yy
> YY FLOAT = 35.0000
> IDL> help,last_row
> LAST_ROW FLOAT = 89.0000
> IDL> print,last_row-yy
> 54.0000
> IDL> print,fix(last_row-yy)
> 53
> IDL> print,fix(last_row)-fix(yy)
> 54
>
> Would appreciate any insight.
>
> Paritosh
It may be because FIX truncates rather than rounds
to get an integer, so FIX(4.9) returns 4 and not 5.
So, as an example, if yy is actually 35.0000000001 and
last_row is 89.0000000000, when you subtract
them and then FIX you get FIX( 53.9999999999) which
is 53. If you FIX them first, you get 89 - 53 which
gives 54.
So the answer may differ depending on how accurately
your machine stores float variables. I think this
is a problem in other languages besides IDL. Try
using ROUND instead of FIX, as this rounds to the nearest
integer, and it should solve the problem.
Tim
|
|
|
|
| Re: BUG in IDL 4.0a PowerMac [message #6040 is a reply to message #6038] |
Fri, 05 April 1996 00:00 |
Andy Loughe
Messages: 174
Registered: November 1995
|
Senior Member |
|
|
Paritosh Dhawale wrote:
>
> Hello,
>
> I get this dangerous bug rather randomly running
> IDL 4.0a on a PowerMac 7500. Any one with similar
> problems? IDL fails to return the correct answer
> in statement 4.
>
> IDL> help,yy
> YY FLOAT = 35.0000
> IDL> help,last_row
> LAST_ROW FLOAT = 89.0000
> IDL> print,last_row-yy
> 54.0000
> IDL> print,fix(last_row-yy)
> 53
> IDL> print,fix(last_row)-fix(yy)
> 54
Ouch!
FYI: I *don't* get this error running IDL 4.01b under Solaris.
--
Andrew F. Loughe [afl@cdc.noaa.gov, http://cdc.noaa.gov/~afl]
University of Colorado, CIRES * Campus Box 449 * Boulder, CO 80309
phone: (303) 492-0707 fax: (303) 497-7013
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
