double precision images in 2.15
Tom Duvall wrote on Jan 12, 2011
I have a fits image in double precision with bitpix=-64 that the new iraf 2.15 sees as all zeroes. Iraf 2.14 sees the data fine. An image with this property is available at http://soi.stanford.edu/~duvall/Data/test_double.fits.
Tom
Tom
Mike Fitzpatrick wrote on Jan 12, 2011
Hi Tom,
This is a known problem I'm looking into currently. It only affects FITS files but is severe enough to push up the next patch. Stay tuned.
-Mike
This is a known problem I'm looking into currently. It only affects FITS files but is severe enough to push up the next patch. Stay tuned.
-Mike
Mike Fitzpatrick wrote on Jan 12, 2011
The problem has been fixed for the upcoming patch release. IEEE NaN masking in reading FITS files was being done in 32-bits causing a zero exponent value. The only workaround for the moment is to use single-precision FITS images or IMH format for doubles (remember to reset the 'imtype' variable).
Last post on Jan 12, 2011