loading nmisc with gemini package
Dick Joyce wrote on Jan 04, 2011
Starting with the V2.15 upgrade, I am unable to run gemini IRAF applications, getting an error message:
ERROR: The NMISC package is required but not defined.
ERROR: task `nmisc' not found
"# Package script for the gemtools package"
line 3: gemtools$gemtools.cl
called as: `gemtools ()'
If I load nmisc explicitly in loginuser.cl:
reset nmisc = /iraf/extern/nmisc
task nmisc.pkg = nmisc$nmisc.cl
then I get a complaint that nmisc has no parameter file if I try to load a gemini package (e.g., niri), even though the file nmisc.par exists in /iraf/extern/nmisc:
ERROR: task `nmisc' has no param file
called as: `cl ()'
I was able to use the niri package with V2.14, so this is a recent development.
Thanks,
Dick
ERROR: The NMISC package is required but not defined.
ERROR: task `nmisc' not found
"# Package script for the gemtools package"
line 3: gemtools$gemtools.cl
called as: `gemtools ()'
If I load nmisc explicitly in loginuser.cl:
reset nmisc = /iraf/extern/nmisc
task nmisc.pkg = nmisc$nmisc.cl
then I get a complaint that nmisc has no parameter file if I try to load a gemini package (e.g., niri), even though the file nmisc.par exists in /iraf/extern/nmisc:
ERROR: task `nmisc' has no param file
called as: `cl ()'
I was able to use the niri package with V2.14, so this is a recent development.
Thanks,
Dick
Mike Fitzpatrick wrote on Jan 04, 2011
Basically, it means NMISC is not defined locally. I've fixed this for the main PC server at NOAO, if you still have problems send me an email telling me which machine you're using.
-Mike
-Mike
James Turner wrote on Jan 04, 2011
Hi Dick,
Note that the gemini package hasn't been ported to IRAF 2.15 yet, so even with nmisc working, gemini is very unlikely to work for you. We're recommending that you continue running the gemini package on 2.14.1 for the time being (of course there's no reason why you can't also keep an IRAF 2.15 installation for other purposes). It will likely be several months before we're able to release a 2.15-compatible version (and I expect it will remain 32-bit only then).
Cheers,
James.
Note that the gemini package hasn't been ported to IRAF 2.15 yet, so even with nmisc working, gemini is very unlikely to work for you. We're recommending that you continue running the gemini package on 2.14.1 for the time being (of course there's no reason why you can't also keep an IRAF 2.15 installation for other purposes). It will likely be several months before we're able to release a 2.15-compatible version (and I expect it will remain 32-bit only then).
Cheers,
James.
Mike Fitzpatrick wrote on Jan 04, 2011
Note that the gemini package hasn't been ported to IRAF 2.15 yet, ....
Why is this so hard to grasp ..... IT DOES NOT MATTER if a package has or has not been "ported" to v2.15, the system is perfectly happy to run old v2.14 (or earlier) binaries right alongside the new v2.15 core system code. "Ported to v2.15" is a matter of compilation, not runtime.
As for GEMINI, it also hasn't been shown there's actually a problem using newer packages were there were minor task shuffles, only a statement that this hasn't been tested fully yet (and I maintain that determining whether there is actually a problem, and even 64-bit support, is a trivial thing). [/quote]
James Turner wrote on Jan 04, 2011
Perhaps you're right that the effort will be less than I'm suggesting, but the changes to nhedit will certainly break things and I don't believe we've figured out how to deal with that yet -- IIRC, in the limited time before Christmas, we couldn't see how to release a version of the package that will work under both 2.14 and 2.15, at least not without detecting the IRAF version.
James.
James.
Mike Fitzpatrick wrote on Jan 04, 2011
IIRC, in the limited time before Christmas, we couldn't see how to release a version of the package that will work under both 2.14 and 2.15, at least not without detecting the IRAF version.
Since you're open to releasing a new version of the package: Just change the use of NHEDIT in your code to use the one installed in the system (which, BTW, was there in v2.14 as well) instead of the old FITSUTIL code (i.e. the "NULL" parameter thing).
Or, import the old NHEDIT task into GEMINI and use a local version (while you're at it, the one or two tasks in GMISC used could also be imported to reduce depencies and make things easier for users, would be more worthwhile than just importing NHEDIT). NMISC is only required because your package loading scripts check for it, the tasks it contains are either already in the system (for a while now) or are obsolete.
All of this will work with either v2.14 or v2.15 with no version detection required. An 64-bit support,..... a one-line code change is required to compile.
Dick Joyce wrote on Jan 04, 2011
Thanks, Mike. The various gemini packages now seem to load OK (I did have to remove my nmisc redefinitions from loginuser.cl). Fortunately I had finished reducing our NIRI data from last semester, but we are expecting some 2011A results and I wanted to be ready.
Dick
Dick
travisrector wrote on Jan 04, 2011
Just to add to the conversation...I have a problem with gmos/imcoadd while running gemini 1.9 with IRAF 2.15.1 macintel 64-bit. The output images are a single data value. I do not have the same problem when reducing the same data with IRAF 2.14 and gemini 1.9. So it looks like there are some 64-bit issues here, even though everything appears to run ok (i.e., no crashes).
Mike Fitzpatrick wrote on Jan 04, 2011
Gemini doesn't support v2.15 or 64-bit, and its not clear this is a 64-bit issue since the task is a 2200+ line script calling multiple tasks. The recommended version is v1.10, if that works with v2.14 then it is likely a v1.9 bug, otherwise it might also be a parameter change in v2.15. You can't say for sure it's a 64-bit problem until you find out which task in the script is causing the corruption, however you can rule out a 64-bit problem by seeing whether a 32-bit v2.15 IRAF works but the 64-bit doesn't.
Last post on Jan 04, 2011