more Max OS X woes!
Maurizio, Patrick, everyone,
It appears there are some serious problems getting the latest
'Ifeffit/Athena' running on Max OS X -- both 10.1 and 10.2. The
sympton is gettting messages like
dyld: perl Undefined symbols:
_pgarro_
_pgask_
which means the Ifeffit perl module is no longer able to find the
PGPLOT library. When this happens, older versions of athena are
also broken. My guess is that this is related to the changes in
horae's Makefile.PL for horae-007, but I haven't looked into it
any further.
I believe the workaround is to go back to the Ifeffit source
distribution and reinstall the ifeffit perl module from there:
cd ...../ifeffit_1.0b/wrappers/perl
perl Makefile.PL
make
make install
and then back-tracking to horae-006.tar.gz:
http://feff.phys.washington.edu/~ravel/software/exafs/packages/horae-006.tar...
At least, this worked on Patrick's laptop last week.
It might be enough to simply backtrack to horae-006.
Sorry for all the trouble,
--Matt
On Wed Dec 11, Maurizio Aiello
Dear Matt, I tried to install ona fresh Mac-osX 10.1.4 athena. I followed all steps necessary, (it required ten hours of continuous compilation) for installing everything. I replaced dynamic loader with the patched one, and in fact now I dont' have multiple definition error. But when i run perl -e "use ifeffit" or I run from within an xterm in Xdarwin the command athena the answer is:
[localhost:~] marzia% perl -e "use ifeffit" dyld: perl Undefined symbols: _pgarro_ _pgask_ .....
If I substitute dyld with the original one now the error is (of course) [localhost:~] marzia% athena dyld: perl multiple definitions of symbol _LangExit /Library/Perl/darwin/auto/Tk/Event/Event.bundle definition of _LangExit /Library/Perl/darwin/auto/Tk/Tk.bundle definition of _LangExit
If I run ifeffit directly from command line it seems to work correctly.
Can you help me? Thanks a lot.
P.S. I am a Unix-Network programmer, I'm doing this job for my wife, she is a researcher. I don't have any knowledge on what these programs do, I only compile and debug and so on :-))))
Thanks a lot
Maurizio Aiello
On Wed Dec 11, Patrick Allen
This is entirely FYI: I couldn't resist and got Jaguar 10.2.2 going now. I reinstalled nearly every package I could think of and got very close with Athena and get the message:
dyld: perl Undefined symbols: _G77_getenv_0 _pgarro_ _pgask_ _pgband_ _pgbbuf_ _pgbox_ etc..... Trace/BPT trap
So for the moment, I chickened out and downloaded the whole darn thing onto my Virtual PC partition on OSX. It works and I can play with my PNC-CAT files with the latest programs. I'm looking forward to coming out to Chicago more now. Seasons greetings!
Regarding problems getting ifeffit/athena to communicate with pgplot on Mac OSX: Hmmm.... I'm confused. I got the current version of horae to work just fine on my officemate's Mac. Paul Fons told me that pgplot was placed in /sw/lib/pgplot/ by fink. I put some logic in the horae build script that checks to see if it is running on Darwin and then adds "-L/sw/lib/pgplot/" to the compile flags for the wrapper. Is that not the right place for pgplot? Another question: what is the reponse when you do this at a terminal command line on your OSX machine: perl -e 'use English; print $OSNAME, $/' The logic in the build script assumes that the answer is "darwin". B -- Bruce Ravel ----------------------------------- ravel@phys.washington.edu Code 6134, Building 3, Room 222 Naval Research Laboratory phone: (1) 202 767 5947 Washington DC 20375, USA fax: (1) 202 767 1697 NRL Synchrotron Radiation Consortium (NRL-SRC) Beamlines X11a, X11b, X23b, X24c, U4b National Synchrotron Light Source Brookhaven National Laboratory, Upton, NY 11973 My homepage: http://feff.phys.washington.edu/~ravel EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/
A lot of typing below. I hope it helps some. Briefly, there is *no* problem with the horae and the latest version of all of the dependent packages. The problems mentioned arise from old versions. If this is relevant, read on... Horae works just fine on about four machines here. One is running an older version of the OS 10.1.2 and hence an older version of fink, pgplot (non-updated), and horae. The problems with unsatisfied dylib references are probably due to the fact that on the machine where the error occurs there is an old version of fink or more specifically pgplot installed. As I recall the location of pgplot changed about 3-6 months ago. The initial location /sw/pgplot didn't fink into the grand organizational scheme of fink and it was moved to /sw/lib/pgplot. At the same time, the compile sequence was changed to use frameworks (a sort of bundle of headers, dynamic library, and related info). I don't think there is any point in supporting an older version of fink as it has gone the way of the dodo (if this is what the problem is). I recognized this problem earlier this year (spring, perhaps) and took the easy way out of leaving the old pgplot library in /sw/pgplot to simplify things (all worked fine). When I installed 10.2, as most of fink was broken due to the OS change, I properly rebuilt pgplot so it is using the latest version from fink (/sw/lib/fink) and modified the compile files (with Bruce) to reflect the changes in fink. This means that for new installs all is fine, however, if someone is using an old (unsupported) version of fink, then I can see the problem happening. As the update to the latest version of fink is free (it is open source), I don't see any reason why not to upgrade fink (specifically gnuplot) on the machines in question. There is information on how to update from various version on the fink homepage at http://fink.sourceforge.net (specifically see the documentation links). Thus to make a long story short. To support the older machines the old makefile sequence is necessary (e.g. for those who are running 10.0, 10.1 -- I think the pgplot packages was updated before 10.2 came out so 10.2 fink installs should not require a special makefile). Having said all this, I am running 10.2.2 on all the installations I support here (I will upgrade the person's machine with 10.1.2 soon) and don't have any 10.1 (or earlier) os installed machine to play with. I would suggest to those that have not upgraded to 10.2 to do so as there are many worthwhile improvements. Rough install/update summary (for 10.2): 0. Make sure that you install the latest version of the developer tools that came with 10.2. Fink will *not* work correctly with the older version of the gcc compiler. This is included literally in the box, but is an optional install. 1. make sure you have the latest version of fink installed (see the upgrade matrix at http://fink.sourceforge.net/download/upgrade.php). Note the fink home page states that they are actively improving the upgrade matrix so it will only get easier to understand :) . For those installing fink for the first time, there is a simple binary installer. 2. Do an "update from cvs" (again see the fink documentation -- for those who prefer a pretty graphical interface -- myself included -- there is a wonderful package called FinkCommander at http:finkcommander.sourceforge.net). This will update the compile descriptions and related information for *all* fink packages. The CLI incantation is: "fink selfupdate-cvs" 3. do a "fink update-all". This will download and recompile all packages which have changed to the most recent versions (as of when you did your last "fink selfupdate-cvs". if you are installing fink for the first time you must install the following modules from fink (note that binaries may be available which will *greatly* reduce install time. You can see this at a glance with FinkCommander (I recommend it). In any case, the following modules are necessary (I think this is inclusive, but I have 232 packages installed and it's hard to check -- fink is a wonderful resource -- at this writing there are 1775 packages available for installing). fink install g77 (installs fortran) fink install python (installs python) fink install pgplot fink install xfree86-rootless-threaded fink install windowmaker (a window manager for X) 4. Make sure that you have installed the 10.2 version of perl/Tk (see http://www.lehigh.edu/~sol0/Macintosh/X/ptk/). I can give a copy of the modified perl/Tk sources to Bruce for downloading on the athena homepage if this would be simpler. Note that if you follow the advice on this page and update perl to 5.8.0 there will be a couple of other things you have to do. See the details at http://www.mail-archive.com/fink-devel@lists.sourceforge.net/ msg02454.html. Briefly perl 5.6 and 5.8 are not binary compatible. This means that after you recompile and replace Apple's ancient perl 5.6 you will have to seek out and destroy (ok rebuild) all binaries that were compiled under 5.6 (e.g. XS modules). This is documented on the link I gave above. Note that some of the perl files in Apple's domain /Library/Perl and /System/Library/Perl will have to be recompiled as well. I don't *think* there is any problem using perl 5.6 with perl/Tk, but on the other hand I haven't tried it so can be sure (I upgraded to perl 5.8 about a week after it came out). Just to be sure 5. Now perl "Makefile.PL" in the horae folder , followed by a "make", followed by a "sudo make install". possible problems include having an non 10.2 compatible version of xfree86 (see the xfree86 on OS X homepage at http://mrcla.com/XonX/). You must use the updater on this site to eliminate some 10.2 related problems. On Thursday, December 12, 2002, at 09:33 AM, Bruce Ravel wrote:
Regarding problems getting ifeffit/athena to communicate with pgplot on Mac OSX:
Hmmm.... I'm confused. I got the current version of horae to work just fine on my officemate's Mac. Paul Fons told me that pgplot was placed in /sw/lib/pgplot/ by fink. I put some logic in the horae build script that checks to see if it is running on Darwin and then adds "-L/sw/lib/pgplot/" to the compile flags for the wrapper.
Is that not the right place for pgplot?
Another question: what is the reponse when you do this at a terminal command line on your OSX machine: perl -e 'use English; print $OSNAME, $/' The logic in the build script assumes that the answer is "darwin".
B
-- Bruce Ravel ----------------------------------- ravel@phys.washington.edu Code 6134, Building 3, Room 222 Naval Research Laboratory phone: (1) 202 767 5947 Washington DC 20375, USA fax: (1) 202 767 1697
NRL Synchrotron Radiation Consortium (NRL-SRC) Beamlines X11a, X11b, X23b, X24c, U4b National Synchrotron Light Source Brookhaven National Laboratory, Upton, NY 11973
My homepage: http://feff.phys.washington.edu/~ravel EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/ _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Paul Fons, Ph.D. Senior Staff Researcher Photonics Institute National Institute for Advanced Industrial Science & Technology Tsukuba, JAPAN 305-8568 email: paul-fons@aist.go.jp tel. +81-298-61-5636 fax: +81-298-61-5615 The lines below are in Japanese 主任研究官 ポール・フォンス 産業技術総合研究所 光技術研究部門 つくば市茨城県 305-8568
Paul, Matt, and others, It seems from what Paul said that the problem is a mixture of my new build scripts and old installations of fink et al. Would it help to put "-L/sw/lib/pgplot/ -L/sw/pgplot" in the compile flags? B -- Bruce Ravel ----------------------------------- ravel@phys.washington.edu Code 6134, Building 3, Room 222 Naval Research Laboratory phone: (1) 202 767 5947 Washington DC 20375, USA fax: (1) 202 767 1697 NRL Synchrotron Radiation Consortium (NRL-SRC) Beamlines X11a, X11b, X23b, X24c, U4b National Synchrotron Light Source Brookhaven National Laboratory, Upton, NY 11973 My homepage: http://feff.phys.washington.edu/~ravel EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/
participants (3)
-
Bruce Ravel
-
Matt Newville
-
Paul Fons