iraf.net

Hello,I am trying to install Frank Valdes' "aceold" onto my Mac OS X computer. I have set arch = generic, and am trying to build the package executables. Halfway through the command
ecl> mkpkg -p aceold
I encounter the message
ld: library not found for -ltbtables
and all compilation stops.I do have Tables installed, and the library is located at
/iraf/extern/tables/lib/libtbtables.a
Since mkpkg seems to find other libraries, such as
/iraf/iraf/lib/libgsurfit.a
with no problem, I tried linking the tbtables library in the /iraf/iraf/lib/ directory. However, mkpkg still stops at the same place for the same reason. Is there anything I am missing?Thanks for your help,
Reiko Nakajima

Two problems: To use the TABLES libraries you need to compile using the '-p tables' flag to import the tables environment, i.e. use the command:[code:1:f119d6ca50] mkpkg -p aceold -p tables [/code:1:f119d6ca50]The second problem is that the $iraf and $IRAFARCH environment variables must be set in your host environment. Since you're compiling inside of IRAF this means in your .cshrc/.bashrc file, or you can simply compile outside of the CL as something like:[code:1:f119d6ca50]
% cd /path/aceold # go to package directory
% setenv aceold /path/aceold # in case you didn't edit extern.pkg
% setenv iraf /iraf/iraf/ # trailing '/' required, must be correct
% setenv IRAFARCH macosx # or 'macintel' for Intel system
% mkpkg -p aceold macosx # set package architecture
% mkpkg -p aceold -p tables # build it
[/code:1:f119d6ca50] The README file with the package appears to leave out the '-p tables' flag, but see the rest for the environment information.
Cheers,
-Mike

Hello Mike,Thanks for your reply. I've tried your suggestions, and here is the error message I get (I've used macintel instead of macosx for my architecture):% mkpkg -p aceold macintel
envinit: cannot open `/iraf/extern/aceoldlib/zzsetenv.def'
Warning, mkpkg line 0: cannot open `aceold$lib/mkpkg.inc'
Warning, mkpkg line 128: could not find mkpkg module or label `macintel'
Warning, mkpkg line 128: no entry in mkpkg file for `macintel'The aceold package doesn't have an aceoldlib/ directory.
% pwd
/iraf/extern/aceold
% ls
total 584
16 README 8 bin@ 0 bin.ssun/
8 aceold.cl 0 bin.freebsd/ 0 doc/
8 aceold.hd 0 bin.generic/ 0 lib/
8 aceold.men 0 bin.linux/ 8 mkpkg
8 aceold.par 0 bin.redhat/ 0 src/
520 aceold.tar 0 bin.sparc/But the zzsetenv.def file is in the lib/ directory. How can I get around this? I'm not really sure where mkpkg is trying to look for aceoldlib/.Thanks in advance,
Reiko

Different two problems:1) My message left off the trailing '/' in the definition of the 'aceold' package path, i.e. it should read[code:1:becd1bf192]setenv aceold /path/aceold/[/code:1:becd1bf192]The 'aceoldlib' is coming from the missing '/' and trying to resolve the path aceold$lib internally. In general, a trailing '/' on pathnames is recommended and won't hurt.2) The toplevel mkpkg apparently doesn't have an entry for macintel. You should edit the aceold$mkpkg file and add the following at the bottom:[code:1:becd1bf192]
macintel: # install Mac OSX Intel binaries
$verbose off
$set DIRS = "lib src"
!$(hlib)/mkfloat.csh macintel -d $(DIRS)
;
[/code:1:becd1bf192]You'll also need to manually create a 'bin.macintel" directory in the /path/aceold directory, but then a "mkpkg -p aceold macintel" should work. In the end, the 'bin' link should be pointing to the 'bin.macintel" directory (bin.generic should always be empty).
-Mike

Thanks again! I've managed to have mkpkg recognize macintel. However, I'm back to the same problem:% mkpkg -p aceold macintel
deleting any dreg .e files left lying about in the source directories... done
archiving and deleting generic objects... done
compressing bin.generic/OBJS.arc [1] 1928
restoring archived macintel objects... no object archive found; full sysgen will be needed% mkpkg -p aceold -p tables
warning: library `libpkg.a' not found
xc -c -w -p aceold -p tables aceoverlay.x bndry.x catio.x convolve.x detect.x evaluate.x filter.x grow.x omwrite.x sky.x skyio.x split.x t_detect.x t_acedisplay.x maskcolor.x
aceoverlay.x:
overlay:
[... many aceold filenames deleted ...]
maskcolor.f:
mcolos:
mcolor:
ar r /iraf/extern/aceold/src/libpkg.a aceoverlay.o bndry.o catio.o convolve.o detect.o evaluate.o filter.o grow.o omwrite.o sky.o skyio.o split.o t_detect.o t_acedisplay.o maskcolor.o
ar: creating archive /iraf/extern/aceold/src/libpkg.a
ranlib libpkg.a
Updated 15 files in libpkg.a
xc -c -w -p aceold -p tables x_ace.x
x_ace.x:
sys_runtask:
x_ace.f:
sysruk:
xc -Nz -p aceold -p tables x_ace.o libpkg.a -lds -lgsurfit -lcurfit -lxtools -ltbtables -o xx_ace.e
[b:2d54e48965]ld: library not found for -ltbtables[/b:2d54e48965]
collect2: ld returned 1 exit status
Warning, mkpkg line 7: module `relink' not found or returned error
move `xx_ace.e' to `aceoldbin$x_ace.e'
$move: file `xx_ace.e' not found
Warning, mkpkg line 19: error moving file xx_ace.e
Warning, mkpkg line 8: module `install' not found or returned error
Warning, mkpkg line 7: module `update@src' not found or returned error
Warning, mkpkg line 3: module `update' not found or returned errorWould you have any suggestions? Thanks again in advance,
Reiko

Is the TABLES package actually installed (i.e. is it declared in the hlib$extern.pkg file, or do you have a 'tables' environment variable set to be path (w/ trailing slash) to the package in your unix environment) or simply unpacked on the machine? Note the /iraf/extern/tables/lib/libtbtables.a should normally be just a link, you'll need to have the 'macintel' binaries for TABLES also installed, the real libtbtables.a should be in tables$bin.macintel/libtbtables.a
-Mike

Hi Mike,It worked! Thank you very much.You were right--it appears that I hadn't installed tables correctly (or at all--I can't remember).The hlib$extern.pkg contained the necessary reset and task lines. The link to the *.a files are also there, but the actual binary files were not there. I went to stsci and put the *.a and *.e binaries in the tables/bin.macintel directory, and then the mkpkg on aceold worked.I now have a separate problem: the links to the *.a binaries in tables is still messed up, and "mkpkg macintel" doesn't seem to work. The tables command in cl> seems to work just fine, though. Should I worry about it?% mkpkg macintel
deleting any dreg .e files left lying about in the source directories... _TFL.mkfloat: Permission denied.
grep: _TFL.mkfloat: No such file or directory
tee: _.e_files: Permission denied
cat: _.e_files: No such file or directory
_DFL.mkfloat: Permission denied.
rm: _TFL.mkfloat: No such file or directory
done
archiving and deleting generic objects... cat: _DFL.mkfloat: No such file or directory
tar: Cowardly refusing to create an empty archive
Try `tar --help' or `tar --usage' for more information.
_TFL.mkfloat: Permission denied.
tar: bin.generic/OBJS.arc: Cannot open: No such file or directory
tar: Error is not recoverable: exiting now
Error: cannot archive generic objects
diff: _DFL.mkfloat: No such file or directory
diff: _TFL.mkfloat: No such file or directory
rm: _DFL.mkfloat: No such file or directory
rm: _TFL.mkfloat: No such file or directory
rm: bin.generic/OBJS.arc: No such file or directorySorry to bug you again.Thanks in advance,
Reiko
[/b]

The mkpkg command still needs the '-p tables' flags to work properly, but the problem appears to be the 'permission denied' message. I'm guessing the files in the package tree are owned by e.g. 'iraf' and you're trying to do the build as yourself. If you installed pre-built binaries there's no need to reset the architecture (the '-p tables' when you build other packages should find the binaries okay, the synlink is used when building TABLES itself).
-Mike

Ah, thanks again! the
% mkpkg -p tables macintelstill gave the same permission error (I am logged in as iraf), so I'm just going to manually add links to the right places. (I'll call again if this causes problems...)Thanks for all your help,
Reiko[/b]

Hello,I've finally managed to run aceold:diffdetect, only to encounter a segmentation violation error:aceold> diffdetect
List of images (mos242s):
List of output object masks (diff242s):
List of output catalogs (diff242s):
ACE:
Image: mos242s - r - DelRA = 0.0, DelDec = 0.0
Bad pixel mask: mos242_bpm
Reference image: mos242s.fits - r - DelRA = 0.0, DelDec = 0.0
Set sky and sky sigma: sky = SKYFIT, sigma = SKYSIG
Fit sky:
start line = 1, end line = 8221, step = 83.0
skyxorder = 4, skyyorder = 4, skyxterms = half
skyhclip = 3., skylclip = 3.
Set sky and sky sigma: sky = SKYFIT, sigma = SKYSIG
Fit sky:
start line = 1, end line = 8221, step = 83.0
skyxorder = 4, skyyorder = 4, skyxterms = half
skyhclip = 3., skylclip = 3.
Detect objects:
Convolution:
1. 1. 1.
1. 1. 1.
1. 1. 1.
ERROR: segmentation violationNo output were produced (both a fits file and a table are supposed to result from diffdetect).(a) Is there any way I can fix this problem? I have no clue as to where to look. (I am vaguely suspecting that my improper installation of the Tables package is causing problems--in which case, I would need a pointer, since the STScI site didn't help me much.)(b) Although it would be nice, I don't necessarily need diffdetect, but I need to be able to run a script named "satzap.cl" which apparently relies on the ACE:diffdetect output -- i.e., "satzap.cl" will not run on a fits file that hasn't gone through diffdetect. What does diffdetect do (probably changes a header item, but which one, and to what?)Thanks for your help,
Reiko

A segfault is usually an error in the code, although it's one that can sometimes be data-dependent. Without being able to reproduce the problem using the data or parameters there's not much I can say, Frank is away this week (ACE is his package) but may comment when he returns if it looks like a familiar problem.Something that might help is a reference to the use of these tasks in the reduction of the NDWFS data at http://www.noao.edu/noao/noaodeep/ReductionOpt/makeStack.html ACE itself hasn't been released as a final package although there has been some work on it recently. When Frank returns he might be able to point you to a more recent version and will likely request more information (such as the data/params) in order to examine the problem. You can post the image and a 'dpar diffdetect' of the params to ftp://iraf.noao.edu/pub with a reply saying where they are and somebody will get back to you if/when the problem if tracked down.Note that I do NOT think this is related to the tables/aceold build issues you were having earlier, but packages labeled 'old' in any way can't be considered foolproof. Post any useful information as you find it.
Cheers,
-Mike

Hi Mike,Thanks again for your reply. I've been looking at the NDWFS reduction webpage up to this point. I think I will have to find a workaround for creating manual masks.I've posted the files
mos242s.fits
mos242_bpm.fits
mos242_diffdetect.txt
to ftp://iraf.noao.edu/pub in case Frank is interested. Thanks for all your help,
Reiko

Hello Reiko,I'm sorry for the difficulties you had. Thank you for the data.The bpm file seems to be multiextension with two images that are all zero. I copied an extension to mos242_bpm.pl. I then ran your example on a redhat machine and it did not fail. However, I see one conceptual problem. You have the reference image being the same as the input image. Without scaling it naturally finds zero sources. With the scaling I got 18 sources while you got a segmentation error on macintel. Before proceeding further, if you are interested, you should use diffdetect as intended where the reference image is actually a different "reference" image. This is usually a stack of a dithered set of exposures.Yours,
Frank ValdesP.S. As you realize and Mike mentioned, aceold is quite old but kept available to be consistent with the NDWFS data reduction guide which is a good reference for this kind of processing.