problem with scopy on a Mac?
phil_hodge wrote on Mar 29, 2010
The tomultispec script in stsdas fails with a bus error when run on a Mac, but it works when run on Linux. The bus error is in the call to scopy (which calls sarith), and it can be reproduced by calling scopy separately on a temporary file created by tomultispec. I can give you a copy of this file if it will help. The file is in FITS format and contains a single 1-D image of length 1024, with average values around 3.e-9, and there are multispec-type keywords defining the wavelengths. I don't see anything peculiar about the image data, so perhaps there's a bug in the way tomultispec has assigned the multispec keywords.
Phil
Phil
Mike Fitzpatrick wrote on Mar 29, 2010
Hi Phil,
Could you please upload the file to the anonftp on iraf.noao.edu:/pub so we can reproduce it? I assume this is a v2.14.1 Intel system we're talking about?
-Mike
Could you please upload the file to the anonftp on iraf.noao.edu:/pub so we can reproduce it? I assume this is a v2.14.1 Intel system we're talking about?
-Mike
Mike Fitzpatrick wrote on Mar 29, 2010
Phil,
Got the file, but what am I supposed to do with it? The help for TOMULTISPEC (in CTOOLS, right?) is expecting a 3-d BINTABLE, not a 1-d spectrum.
-Mike
Got the file, but what am I supposed to do with it? The help for TOMULTISPEC (in CTOOLS, right?) is expecting a 3-d BINTABLE, not a 1-d spectrum.
-Mike
phil_hodge wrote on Mar 29, 2010
fitz
Phil,
Got the file, but what am I supposed to do with it? The help for TOMULTISPEC (in CTOOLS, right?) is expecting a 3-d BINTABLE, not a 1-d spectrum.
-Mike
Run scopy with tomul_Y_1176f.fits as input, o8yg.imh as output, and renumber=yes. I wouldn't expect you to debug tomultispec (yes, it's in ctools); the problem we're seeing is in scopy, which is called by tomultispec. Other than input, output and renumber, all the parameters for scopy have their default values.
Thanks.
Phil
Mike Fitzpatrick wrote on Mar 29, 2010
Phil,
No joy, works just fine for me (on two different machines):
All I can suggest is that you be sure to have all parameters unlearned to we're running the same test, e.g. do a fresh MKIRAF in /tmp and login there. Nothing's changed in the task for a while and there are no known bugs.
-Mike
No joy, works just fine for me (on two different machines):
onedspec> scopy tomul_Y_1176f.fits o8yg.imh renumber+
onedspec>
All I can suggest is that you be sure to have all parameters unlearned to we're running the same test, e.g. do a fresh MKIRAF in /tmp and login there. Nothing's changed in the task for a while and there are no known bugs.
-Mike
phil_hodge wrote on Mar 29, 2010
The problem wasn't with scopy at all. The user had imdir set to a directory that didn't exist, in a directory that was not writable. Here is an example, run on a Linux machine:
ecl> set imdir = "/iraf/imdirs/hodge/"
ecl> noao
ecl> onedspec
onedspec> scopy tomul_Y_1176f.fits o8yg.imh renumber+
ERROR: segmentation violation
"verbose=verbose)"
line 29: onedspec$scopy.cl
called as: `scopy (input=tomul_Y_1176f.fits, output=o8yg.imh, renumber=yes)'
I set imdir to "HDR$", so I hadn't encountered this problem before.
Phil
ecl> set imdir = "/iraf/imdirs/hodge/"
ecl> noao
ecl> onedspec
onedspec> scopy tomul_Y_1176f.fits o8yg.imh renumber+
ERROR: segmentation violation
"verbose=verbose)"
line 29: onedspec$scopy.cl
called as: `scopy (input=tomul_Y_1176f.fits, output=o8yg.imh, renumber=yes)'
I set imdir to "HDR$", so I hadn't encountered this problem before.
Phil
Last post on Mar 29, 2010