massey |
01/15/2008 03:22PM (Read 6869 times)
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
I've just installed the finder package on my intel Mac. However, when I go to run "tfinder" using the standard GSC CDs, I get the following error message:
ERROR: can't open table, it appears to be byte-swapped
"gscfind (ira, idec, iequinox, width, > tmp1)"
line 172: finder$src/tfinder.clNow, this works fine on my old G5 Mac, and indeed the bytes are swapped on the Intel relative to the G5. Is this a problem with the CDs, or a problem with some internal table?
|
|
|
|
fitz |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
Hi Phil,The tables used by FINDER are machine specific but I think all you need to do is run MKGSCINDEX to recreate them for the local machine. If you haven't got the GSC CD's let me know, I think I can dig up an intel-ordered table you can drop in.-Mike
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
I'm afraid it's proving worse than that. I've emailed Rob directly. Here's what I'm seeing:I ran mkgscindex on the "north" cd, and it seemed to work:finder> mkgscindex
GSC index FITS file (/cdrom0/tables/regions.tbl): gscnorth$tables/regions.tbl
output table root pathname (finder$lib/index):
index file finder$lib/index already exists, overwrite? (yes):
Fits_file IRAFNAME Dimensions BP DATE OBJECT
gscnorth$tables/ finder$lib/index.fi 8 01/06/89
regions.tbl ts I then went to run it on the south:
mkgscindex
GSC index FITS file (gscnorth$tables/regions.tbl): gscsouth$tables/regions.tbl
output table root pathname (finder$lib/index):
Fits_file IRAFNAME Dimensions BP DATE OBJECT
gscsouth$tables/
Warning: Cannot rename finder$lib/indexa.fits to finder$lib/index.fits
finder$lib/indexa.f 8 01/06/89
regions.tbl its
Well, anyway, I went to run tfinder (on a southern field) and got the following:
Searching the Guide Star Catalog index...
ERROR: Table `/scisoft/all/packages/iraf/extern/finder/lib/index.tab' does not exist or cannot be opened.
"gscfind (ira, idec, iequinox, width, > tmp1)"
line 172: finder$src/tfinder.cl
called as: `tfinder ()'Actually I remember this as being a problem I've had before. What is in there now is "indexa.fits", not
"indexa.tab", and similarly "index.fits" and not "index.tab".So, I just did a "mv index.fits index.tab".It then craps out as follows:
Searching the Guide Star Catalog index...Reading the Guide Star Catalog regions:
gscsouth$gsc/s6000/8959.gsc
ERROR: Parameter not a legal character constant (datatype)
"oldirafname=yes, offset=0, > tmp2)"
line 181: finder$src/tfinder.cl
called as: `tfinder (mode=h)'
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
Rob suggested the following I simply re-run mkgscindex. Unfortunately, that didn't work: it generated a finder$lib/index.fits rather than an index.tab.Nothing I have done (set imtype='imh', set imtype='hhh') has fixed that, and
just renaming the index.fits to an index.tab doesn't work either.So, I'm stuck.Mike---I'll take you up on your generous offer of a correctly byte-swapped copy
of the tables...
|
|
|
|
fitz |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
Phil,See https://iraf.net/ftp/pub/fitz/index.tab.SSUN
https://iraf.net/ftp/pub/fitz/index.tab.LNUXThese *should* work for the G5 and Intel respectively.Cheers,
-Mike
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
Hi, Mike---I downloaded the file and placed it in finder$lib as "index.tab". But, I'm
afraid it doesn't work. When I go to run tfinder, I get this error message:
Searching the Guide Star Catalog index...
ERROR: Could not find column REG_NO
"gscfind (ira, idec, iequinox, width, > tmp1)"
line 172: finder$src/tfinder.cl
called as: `tfinder ()'If I run a tprint on it, the file can be read, but it looks funny. (OK, what do I mean by "funny"? I mean it looks like this:
# Table index.tab Thu 08:08:29 17-Jan-2008# row _GER H_ARWOL_n
# _GE H_A 1 1 0
2 2 0
3 3 0
4 4 0
5 5 0
6 6 0
7 7 0
8 8 0etc. Sorry this is proving to be such a pain.Any idea why I can't get "mkgscindex" to generate a index.tab
file rather than an "index.fits" file? I tried Rob's suggesting of
explictly doing a "set imtype='imh'" and, when that didn't work,
a "set imtype='hhh'". But, no matter what I get the @#*@ .fits
rather than a ".tab". I seem to recall having this problem before...thanks,
phil
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
Hi, Mike et al.I've now successfully generated a new index.tab table following you suggestion of doing a tcopy of index.fits to index.tab. I've compared this
new index.tab (on my Intel machine) to one that works on my G5 Mac.
Let me add here that the Intel machine is running IRAF 2.13-BETA2
while the G5 Mac is running 2.12-2-EXPORT. The finder version
is identical on the two, v2.2, 2000-02-11, at least according to what
an lpar finder says. The problem is that tfinder still doesn't WORK on my Intel Mac. This is what happens when I try to do a solution for an image:
finder> set gscsouth="/Volumes/USA_AURA_STSI_GSC1_1002/"
finder> dir gscsouth$
GSC README.TXT TABLES
(so we know it is reading the CD)
Image name (N3603V3.fits):
Log file N3603V3.fits.log exists, overwrite the file? (no): yesSearching the Guide Star Catalog index...Reading the Guide Star Catalog regions:
gscsouth$gsc/s6000/8959.gsc
ERROR: Parameter not a legal character constant (datatype)
"oldirafname=yes, offset=0, > tmp2)"
line 181: finder$src/tfinder.cl
called as: `tfinder ()'
Here are the parameters of tfinder:
finder> lpar tfinder
image = "N3603V3.fits" Image name
(rootname = "") Alternate root name for output files
(objects = "") List of program object X,Y coords\n
(scale = 0.435) Plate or image scale ("/pixel)
(north = "bottom") Direction of North in the field
(east = "left") Direction of East in the field\n
(ra = 11.252175555556) RA of the reference point (hours)
(dec = -61.254952777778) Dec of the reference point (degrees)
(equinox = 2000.) Reference coordinate equinox\n
(xref = 605.) X coordinate of the reference point
(yref = 532.) Y coordinate of the reference point
(date_obs = "2007-11-15") Date of the observation (YYYY-MM-DD)\n
(update = yes) Update image header WCS following fit?
(interactive = yes) Enter interactive image cursor loop?
(autocenter = no) Center at the catalog coords when entering task
(reselect = no) Apply selectpars when entering task?
(autodisplay = yes) Redisplay after all-source keystroke command?
(verbose = yes) Print a running commentary?\n
(rotate = 0.) Relative position angle (CCW positive)
(boxsize = 9) Centering box full width
(edge = 200.) Edge buffer width (pixels)\n
(opaxis = no) Is the reference point on the optical axis?
(del_ra = 0.) RA offset of the field center (degrees)
(del_dec = 0.) Dec offset of the field center (degrees)\n
(list = "")
(mode = "ql") Now, if I then copy the image over to my G5 machine, and try it there... well, you know where this is going, don't you?It works fine:
Searching the Guide Star Catalog index...Reading the Guide Star Catalog regions:
gscsouth$gsc/s6000/8959.gscExtracting overlapping sources from regions:
GSC_REGION_08959.tabInteractive centering using TPEAK:
The size of the markers matches the centering box (9 pixels).
Change the size with the command `:eparam tvmark'.So, the three possibilities I see are the following:
It's a version thing---either tfinder really is different versions or it's
an issue of 2.12 (working) vs 2.13 (not working).It's an G5 (working) vs Intel (not working)There's still something screwy with my index.tab file. But, here's the
first few lines of tprint from each machine:
G5 (works):
# Table index.tab Thu 11:29:52 23-Mar-2006# row REG_NO RA_H_LOW RA_M_LOW RA_S_LOW RA_H_HI RA_M_HI RA_S_HI DECSI_LO
# 1 1 0 0 0.00 0 10 0.00 +
2 2 0 10 0.00 0 20 0.00 +
3 3 0 20 0.00 0 30 0.00 +
4 4 0 0 0.00 0 10 0.00 +
5 5 0 10 0.00 0 20 0.00 +
6 6 0 20 0.00 0 30 0.00 +
7 7 0 0 0.00 0 10 0.00 +
8 8 0 10 0.00 0 20 0.00 +
9 9 0 20 0.00 0 30 0.00 +
10 10 0 30 0.00 0 40 0.00 +
11 11 0 40 0.00 0 50 0.00 +
12 12 0 50 0.00 1 0 0.00 +
13 13 0 30 0.00 0 40 0.00 +
14 14 0 40 0.00 0 50 0.00 +
15 15 0 50 0.00 1 0 0.00 +
16 16 0 30 0.00 0 40 0.00 +
17 17 0 40 0.00 0 50 0.00 +
18 18 0 50 0.00 1 0 0.00 +From the intel machine (doesn't work):
# Table index.tab Sat 10:00:08 19-Jan-2008# row REG_NO RA_H_LOW RA_M_LOW RA_S_LOW RA_H_HI RA_M_HI RA_S_HI DECSI_LO
# 1 1 0 0 0.00 0 10 0.00 +
2 2 0 10 0.00 0 20 0.00 +
3 3 0 20 0.00 0 30 0.00 +
4 4 0 0 0.00 0 10 0.00 +
5 5 0 10 0.00 0 20 0.00 +
6 6 0 20 0.00 0 30 0.00 +
7 7 0 0 0.00 0 10 0.00 +
8 8 0 10 0.00 0 20 0.00 +
9 9 0 20 0.00 0 30 0.00 +
10 10 0 30 0.00 0 40 0.00 +
11 11 0 40 0.00 0 50 0.00 +
12 12 0 50 0.00 1 0 0.00 +
13 13 0 30 0.00 0 40 0.00 +
14 14 0 40 0.00 0 50 0.00 +
15 15 0 50 0.00 1 0 0.00 +
16 16 0 30 0.00 0 40 0.00 +
17 17 0 40 0.00 0 50 0.00 +
18 18 0 50 0.00 1 0 0.00 +Suggestions? I can keep running on the G5 for a while, but I've moved the machine to another building for student use, and I'd
like to get this working on the new (well, now 6 month old) Intel
machine.Sorry this is proving to be a bit of a bug a boo.thanks!
---phil
|
|
|
|
fitz |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
Phil,Starting from first principles, the 'datatype' parameter is a parameter of the TCALC task used by the scripts, so I'd try unlearning that on the bad machine (or checking whether a login from a fresh MKIRAF in a scratch directory makes a difference). This same problem was reported several times in 2002, however the only thing in the sitemail from Rob is a mention that "a fix was installed on ursa" but no specifics. There is a v2.2.1 of FINDER from Mar06 that might contain the fix, I think that's also the version in the download area on this site.Cheers,
-Mike
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
Hi, Mike---I tried your suggestion---did a mkiraf in a new directory and tried running tfinder from there on the Intel machine but same problem.However, I think I've identified the problem. I think there's something wrong with the version of finder being distributed from the iraf.net site. Here's why:I went to the machine where tfinder DOES work, and moved the finder
external package to a safe place. I then reinstalled finder using the version macosx binaries that are linked from iraf.net. (I also made a new
directory and a new mkiraf just so everything would be pure.) tfinder now fails on that Mac as well. (Admittedly the error message is different: it gives me a segmentation fault.) Now, the binaries are different of course---in one case the macosx and the other macintel---but maybe they were made from a source code with the same problem?The version of finder that DOES work on the macosx system was one that I downloaded several years ago, and probably got it directly from the NOAO site. Or maybe Rob gave me a "fixed" version, I dunno---I remember I had some problems when I first went to use it on the Mac.Anyway, my prediction is that if you put the iraf.net version of finder on a mac, you'll see it fail too. If that's not true then we're back at square one.thanks,
phil
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
I think this also explains the other mystery from last week, why finder was working on my G5 Mac but doing the exact same thing on my G4 Mac didn't work. The version on the G4 mac was a fresh install from the IRAF site.
|
|
|
|
paco |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/24/2006
Posts: 26
|
Hi Phil, Mike, et al.,
Did this problem with tfinder on Intel Macs get sorted out? I was about to try to install it for the first time on mine, and downloaded the macosx binaries from the noao website. But before I go through the hassle that you have had, Phil, I'd like to know if the current version does in fact work. Otherwise, I'll go about my astrometry in the somewhat more laborious way that I know does at least work.thanks,
Frank
|
|
|
|
massey |
01/15/2008 03:22PM
|
|
|
Status: offline
Registered: 02/10/2006
Posts: 162
|
No, I'm afraid that I'm still waiting. The problem turns out not to be Intel
vs non-intel---if I download the "new" tfinder onto my G5 machine I have the
identical problem. An older version of the binary works. So, I think this may
be a version control issue.It may also be that Rob fixed this problem for me at one point on my G5, and
that I've been running with my own customized version, but that the fix
never propigated back to the main version.However, it does work fine if you generate your own table of standard
coordinates. This is actually what I've been doing, since it means I can
use the ucac2. The new GSS (which works on my "old" version) is almost
as good, so I do think it would be if this got resolved.
|
|
|
|