iraf.net

I use a vector subroutine aupxr to unpack a series of complex numbers into real and imaginary parts. When the input number is complex(0.,0.), the output is 128. for both the real and imaginary parts. But only iraf v15 sees this. Iraf v14 (both on linux) sees the correct thing. It doesn't seem to matter if I do the conversion on v14 or v15, the output is seen as wrong on v15 and correct on v14.I made an output fits image that shows this. It is on http://soi.stanford.edu/~duvall/Iraf/ctestr.fits. It is a 1-d image with the first half of the line should be zero. Listpixels shows the problem. Iraf v15 really sees the pixels as 128 as one can find by squaring the image.
Tom

The good news is that it appears the FITS file actually has zero values on disk, the bad news is that it appears to be similar to another recent report in which reading real zero values from FITS produces the 128 value. A quick test program confirms that the aupxr() routine works as expected.I'm looking at it now, will post back when I get a solution.