problem with SPLOT
giannicatanzaro wrote on Mar 08, 2011
I recently installed version 2.15 64bit and I get this error when I try
to SPLOT an image:
echelle> splot LJHB0004
Image line/aperture to plot (0:) (1):
ERROR: Pixel subscript out of bounds (LJHB0004)
can you give me some help?
thanks
Gianni
to SPLOT an image:
echelle> splot LJHB0004
Image line/aperture to plot (0:) (1):
ERROR: Pixel subscript out of bounds (LJHB0004)
can you give me some help?
thanks
Gianni
Mike Fitzpatrick wrote on Mar 08, 2011
A 'pixel subscript out of bounds' usually means the FITS file is corrupted (isn't properly padded with zeroes to a full record, is scrambled somehow etc), or perhaps has a header problem. If you could post the image to the anonftp at ftp://iraf.noao.edu/pub we can probably spot the problem, otherwise there is no known issue with the task to explain the problem.
Nancy D. Morrison wrote on Mar 08, 2011
I get the same problem with splot, to wit:
ecl> splot 6.fits
Image line/aperture to plot (0:) (5):
ERROR: Pixel subscript out of bounds (6.fits)
The 2.15.1a patch is installed, but does not help.
I am running on Snow Leopard on a 17-inch MBP, 64-bit, recently installed iraf 2.15 in the scisoft installation.
The error does not occur when I plot a wavelength-calibrated spectrum with a single line or with 9 lines (haven't tried a larger number) but only with a raw image with no wavelength calibration (x axis in pixels), in this case 800 X 1200. I can plot the image with implot, but splot used to work in iraf 2.14 in this setting, and splot is sooo much nicer for what I am trying to do.
I have successfully plotted the same image and many others like it with IRAF 2.14, so the image is not corrupted. I'll gladly upload the image if you like, but you might want to try it yourself with a raw image of your own.
ecl> splot 6.fits
Image line/aperture to plot (0:) (5):
ERROR: Pixel subscript out of bounds (6.fits)
The 2.15.1a patch is installed, but does not help.
I am running on Snow Leopard on a 17-inch MBP, 64-bit, recently installed iraf 2.15 in the scisoft installation.
The error does not occur when I plot a wavelength-calibrated spectrum with a single line or with 9 lines (haven't tried a larger number) but only with a raw image with no wavelength calibration (x axis in pixels), in this case 800 X 1200. I can plot the image with implot, but splot used to work in iraf 2.14 in this setting, and splot is sooo much nicer for what I am trying to do.
I have successfully plotted the same image and many others like it with IRAF 2.14, so the image is not corrupted. I'll gladly upload the image if you like, but you might want to try it yourself with a raw image of your own.
Mike Fitzpatrick wrote on Mar 08, 2011
Yes, please upload your problem image.
Note also that scisoft included the original v2.15 release and recent patches have fixed some known issues. You should be sure to install the patch1.macx.x96_64.gz in the iraf root directory to overlay the patch file for OSX 64-bit on your existing version (it will replace binaries). If you still have problems after that, it may be data-related.
Note also that scisoft included the original v2.15 release and recent patches have fixed some known issues. You should be sure to install the patch1.macx.x96_64.gz in the iraf root directory to overlay the patch file for OSX 64-bit on your existing version (it will replace binaries). If you still have problems after that, it may be data-related.
Nancy D. Morrison wrote on Mar 08, 2011
OK, I've uploaded a problem image to ftp://iraf.noao.edu/pub. The image is called splotbad20010409_6s.fits - it's only a small section of one of the large images I described earlier. It gives the same error, showing that the size of the image is not the issue.
When I posted previously, I had already installed the patch1.macx.x96_64.gz. When IRAF starts, it says:
NOAO/IRAF PC-IRAF Revision 2.15.1a EXPORT Mon Feb 21 18:54:16 MST 2011.
The splot problem is not changed by the patch.
Thanks for your attention to this.
When I posted previously, I had already installed the patch1.macx.x96_64.gz. When IRAF starts, it says:
NOAO/IRAF PC-IRAF Revision 2.15.1a EXPORT Mon Feb 21 18:54:16 MST 2011.
The splot problem is not changed by the patch.
Thanks for your attention to this.
Nancy D. Morrison wrote on Mar 08, 2011
Another thought - it occurred to me that the issue might be the datatype of the problem image, which is 'short.' However, I tried to plot a 'real' image with splot, and I got the same error.
ecl> imhead biasave
biasave[800,1200][real]: bias
ecl> splot biasave
Image line/aperture to plot (0:) (9):
ERROR: Pixel subscript out of bounds (biasave)
ecl> imhead biasave
biasave[800,1200][real]: bias
ecl> splot biasave
Image line/aperture to plot (0:) (9):
ERROR: Pixel subscript out of bounds (biasave)
Mike Fitzpatrick wrote on Mar 08, 2011
Thanks for posting the image. The problem was a bug in the 64-bit port of v2.15 that's been around from the initial alpha release but only shows up with certain data. The bug and fix are described in
http://iraf.net/article.php/20110309004941989
and the Buglog #578 story on the main page of this site.
Since multiple binaries are affected, the v2.15.1a patch was re-issued to fix this for future installations and minimize the damage, likewise external packages for this bug were also rebuilt. Please re-install the patch1 file for your system to fix the file and do a "make update" in the $iraf/extern directory to update your packages. If you still have problems please let us know, thanks for the report.
-Mike
http://iraf.net/article.php/20110309004941989
and the Buglog #578 story on the main page of this site.
Since multiple binaries are affected, the v2.15.1a patch was re-issued to fix this for future installations and minimize the damage, likewise external packages for this bug were also rebuilt. Please re-install the patch1 file for your system to fix the file and do a "make update" in the $iraf/extern directory to update your packages. If you still have problems please let us know, thanks for the report.
-Mike
Nancy D. Morrison wrote on Mar 08, 2011
As you suggested, I downloaded and installed the modified version of patch1.macx.x86_64.gz dated 3/9/11 7:07 AM.
Since I have the SciSoft installation, I followed the instructions here. They did succeed in installing the 2.15.1a patch the first time.
http://web.mac.com/npirzkal/Scisoft/News/Entries/2011/1/21_2011.1.1_Beta_1_Released.html
However, it appears that no update was actually made:
1. I still have the ERROR: Pixel subscript out of bounds in splot
2. When IRAF starts, it says,
NOAO/IRAF PC-IRAF Revision 2.15.1a EXPORT Mon Feb 21 18:54:16 MST 2011
(today's date should be here, right?)
3. The output from 'make update' says all packages are up to date.
That patch file is really new, correct? Have I done something stupid? Thanks.
Since I have the SciSoft installation, I followed the instructions here. They did succeed in installing the 2.15.1a patch the first time.
http://web.mac.com/npirzkal/Scisoft/News/Entries/2011/1/21_2011.1.1_Beta_1_Released.html
However, it appears that no update was actually made:
1. I still have the ERROR: Pixel subscript out of bounds in splot
2. When IRAF starts, it says,
NOAO/IRAF PC-IRAF Revision 2.15.1a EXPORT Mon Feb 21 18:54:16 MST 2011
(today's date should be here, right?)
3. The output from 'make update' says all packages are up to date.
That patch file is really new, correct? Have I done something stupid? Thanks.
Mike Fitzpatrick wrote on Mar 08, 2011
I verified the patch file contains the updated binaries, you should likewise do a
to verify that the file was updated on March 8. Otherwise, you might have been in the wrong directory when you unpacked the tar file and are still running the original binary.
cl> dir noaobin$x_onedspec.e long+to verify that the file was updated on March 8. Otherwise, you might have been in the wrong directory when you unpacked the tar file and are still running the original binary.
Nancy D. Morrison wrote on Mar 08, 2011
Thanks for the suggestion. Here is the result. My system time is EDT.
So my binaries _were_ updated? This corresponds to the approximate time I made the latest attempt to update.
ecl> dir noaobin$x_onedspec.e long+
xb-rwr-r- 4807528 Mar 9 16:27 noaobin$x_onedspec.e
Again, no improvement in the error.
So my binaries _were_ updated? This corresponds to the approximate time I made the latest attempt to update.
ecl> dir noaobin$x_onedspec.e long+
xb-rwr-r- 4807528 Mar 9 16:27 noaobin$x_onedspec.e
Again, no improvement in the error.
Mike Fitzpatrick wrote on Mar 08, 2011
That's the right file size for the new binary and using it on my system with your uploaded image works fine. What is the output of the command:
Is there perhaps another x_onedspec.e on the system somewhere?
cl> =osfn("noaobin$")Is there perhaps another x_onedspec.e on the system somewhere?
Nancy D. Morrison wrote on Mar 08, 2011
Thanks for the reply. To your questions:
If I list the other directory I get:
I wonder whether the versions of Sep 8, 2008 are the previous installation (2.14) moved here by the installer when I installed the latest scisoft. IRAF 2.14 worked fine.
For what it's worth, one of my tcsh environment variables is listed as:
PWD=/usr/local/scisoft/packages/iraf/irafbin/noao.bin.macintel
Insight? Thanks!
ecl> =osfn("noaobin$")
/usr/local/scisoft//packages/iraf/iraf/noao/bin.macintel/find / -name x_onedspec.e -print
...
/usr/local/scisoft/packages/iraf/iraf/noao/bin.macintel/x_onedspec.e
...
/usr/local/scisoft/packages/iraf/irafbin/noao.bin.macintel/x_onedspec.e
ecl> dir noaobin$ lo+
-t-r-r-r- 3821 0 Feb 25 2010 noaobin$IRAF.NET
-t-r-r-r- 3821 0 Feb 25 2010 noaobin$NB.MACX.x86_64
-b-rwr-r- iraf 6347189 Feb 21 16:54 noaobin$OBJS.arc.Z
-b-rwr-r- iraf 43928 Feb 21 16:50 noaobin$libasttools.a
-b-rwr-r- iraf 265504 Feb 21 16:50 noaobin$libsmw.a
xb-rwr-r- root 2680736 Mar 9 16:27 noaobin$x_apextract.e
xb-rwr-r- root 3199000 Mar 9 16:27 noaobin$x_apphot.e
...
If I list the other directory I get:
ecl> cd /usr/local/scisoft/packages/iraf/irafbin/noao.bin.macintel/
ecl> dir lo+
-t-rwr-r- 3821 0 Dec 1 2007 IRAF.NET
-t-rwr-r- 3821 0 Dec 1 2007 NB.MACX.X86
-b-rwr-r- 3821 8248297 Sep 8 2008 OBJS.arc.Z
-b-rwr-r- 3821 50892 Sep 8 2008 libasttools.a
-b-rwr-r- 3821 307308 Sep 8 2008 libsmw.a
xb-rwr-r- 3821 1983844 Sep 8 2008 x_apextract.e
xb-rwr-r- 3821 2340236 Sep 8 2008 x_apphot.e
xb-rwr-r- 3821 1906004 Sep 8 2008 x_artdata.e
...
I wonder whether the versions of Sep 8, 2008 are the previous installation (2.14) moved here by the installer when I installed the latest scisoft. IRAF 2.14 worked fine.
For what it's worth, one of my tcsh environment variables is listed as:
PWD=/usr/local/scisoft/packages/iraf/irafbin/noao.bin.macintel
Insight? Thanks!
Mike Fitzpatrick wrote on Mar 08, 2011
If the 2008 binaries were being run then there shouldn't be a problem (at least, not the problem I fixed), but there should only be one set of binaries in the system anyway.
An easy way to see exactly which binary is being run is to start up SPLOT in one window, and then in another terminal use the command
If the path is /usr/local/scisoft/packages/iraf/iraf/noao/bin.macintel/x_onedspec.e
and that file is 4807528 bytes big then it's definitely running the patched binary and there may be a second bug to be fixed.
An easy way to see exactly which binary is being run is to start up SPLOT in one window, and then in another terminal use the command
% ps -efwww | grep onedspecIf the path is /usr/local/scisoft/packages/iraf/iraf/noao/bin.macintel/x_onedspec.e
and that file is 4807528 bytes big then it's definitely running the patched binary and there may be a second bug to be fixed.
giannicatanzaro wrote on Mar 08, 2011
fitz
A 'pixel subscript out of bounds' usually means the FITS file is corrupted (isn't properly padded with zeroes to a full record, is scrambled somehow etc), or perhaps has a header problem. If you could post the image to the anonftp at ftp://iraf.noao.edu/pub we can probably spot the problem, otherwise there is no known issue with the task to explain the problem.
I uploaded the image LJHB0004.fts. I use IRAF v2.15 on a 64bit system based on openSUSE 11.3
thanks for your help
Nancy D. Morrison wrote on Mar 08, 2011
fitz
An easy way to see exactly which binary is being run is to start up SPLOT in one window, and then in another terminal use the command
% ps -efwww | grep onedspec
If the path is /usr/local/scisoft/packages/iraf/iraf/noao/bin.macintel/x_onedspec.e
and that file is 4807528 bytes big then it's definitely running the patched binary and there may be a second bug to be fixed.
Did the test you suggested, your criteria are satisfied, and yes, the patched binary is running. You are right, the old binaries shouldn't be there.
I guess my next step is to try additional onedspec tasks and see what symptoms show up.
An image from our other camera also draws the splot error. I'll do further tests during the next few days and get back to you.
Thanks for all the help! I have learned some Unix tips from this dialog.
Mike Fitzpatrick wrote on Mar 08, 2011
Please try applying the patch file again. I've found a problem that would explain why it works on one machine but not another and rebuilt the binaries after fixing it. For both uploaded test images, and for all my test machines, SPLOT now seems to work as it is supposed to.
Sorry for the inconvenience.
Sorry for the inconvenience.
Nancy D. Morrison wrote on Mar 08, 2011
It works now! Thank you for persevering!
On to the beauties of exploration of the universe!
On to the beauties of exploration of the universe!
Last post on Mar 08, 2011