I wrote earlier that upon trying to install athena on a colleagues powerbook, I ran into problems with an undefined value in a call to the perl standard routine configure. e.g. Tk::Error: Can't call method "configure" on an undefined value at /sw/bin/athena line 1775, <R> line 9. It turns out that the problem was related to an earlier windows-related post (aka the Ravel-ware post). Although I had "cleaned" and reinstalled several times athena and Tk, I hadn't given a second thought to tkatoms -- it had no relation to athena, right? wrong! Upon reintalling tkatoms, athena worked. There must be a perl module that is installed only with tkatoms that is required by athena. Consistent with this is the observation that artemis worked out-of-the-box (e.g. a simple make) without problem. In any case all is solved. It might be a good idea however to investigate this issue further as I am fairly sure that this problem will come up again (not with me though!). In my case (on my own machines), I installed all three programs (tkatoms, athena, and artemis) and never ran into the problem. Paul
Hi Paul,
It turns out that the problem was related to an earlier windows-related post (aka the Ravel-ware post). Although I had "cleaned" and reinstalled several times athena and Tk, I hadn't given a second thought to tkatoms -- it had no relation to athena, right? wrong! Upon reintalling tkatoms, athena worked.
Thanks!! I believe that Bruce is at the NSLS User Meeting this week, so I'll attempt to respond for both of us.
There must be a perl module that is installed only with tkatoms that is required by athena. Consistent with this is the observation that artemis worked out-of-the-box (e.g. a simple make) without problem. In any case all is solved. It might be a good idea however to investigate this issue further as I am fairly sure that this problem will come up again (not with me though!)....
Agreed! The instatllation of Ifeffit + Applications on Unixes (including Mac OS X) does seem complicated and fragile enough (especially with respect to atoms, athena, artemis) that we should think about changing it. I'm open for suggestions, but here's one possible approach: 1. The 'Ifeffit installation' with: ./configure ; make ; make install installs *only* the base library, compiled applications (ifeffit,feffit,autobk,diffkk) and configuration files for the extension languages. These steps do not install the perl/python modules. 2. Basic perl/python wrappers could be installed either with: a) make wrappers ; make wrappers_install immediately after the base library installation b) downloading separate kits for the extensions, copying the local config from /usr/local/share/ifeffit and making according to the scripting language perl Makefile.PL ; make install python setup.py install 3. Applications using the extension languages (or in the case of atoms/athena/artemis, groups of related apps) could be maintained and installed separately from the ifeffit library. For example, atoms/athena/artemis could be installed with: install base Ifeffit download and unpack Ravel_AAA_IfeffitPerl.tar.gz perl Makefile.PL ; make install This might make it a lot easier on Bruce to consolidate, code and interoperability of the A* perl programs. A similar approach could be used for python Apps, hopefully including the new fitting GUI from Sam Webb. At this point, the wrappers and GUIs are changing faster than the base library, and more people are developing applications than the library. Also, it seems possible that there may soon be enough apps that some people may not need them all. Anyway, an approach that allows separate releases of the library and different applications has some appeal. For Win32, an installer that contains the latest version of 'all apps' is needed, of course. A binary package (at least of the base library) for Mac OS X, and possibly even linux, may also be worth considering. This basically means an amicable divorce of the source distribution for the Ifeffit base and Atoms/Athena/Artemis. I think it will actually improve installation on the Unixes. Any opinions on this? Thanks, --Matt
participants (2)
-
Dr. Paul Fons -
Matt Newville