After consulting several Mac users, I decided that there were very few
32-bit Mac operating systems still in use by crystallographers, so (as
stated on the new SHELX homepage) I am only providing a 64-bit
executables for Macs. The Mac executables do not run on old Macs with
PPC processors either. For Linux and Windows I provide both 32 and
64-bit versions. One problem is that even if I tried to cross-compile a
32-bit version for Macs (with -m32 on the ifort compile/link command
line?), I would have no way of testing it, and I would not wish to
distribute untested programs. It might destroy SHELX's reputation for
robustness. All SHELX programs in this release are as far as physically
possible statically linked, which explains why the executables are large
(though the Intel MKL FFT can take some of the blame for that too).
Static linking is an essential part of the SHELX 'zero dependencies'
philosophy.
Apart from some trivial problems like some programs pretending to be
beta-test versions when in fact they are the final release (I had simply
forgotten to change the text), the complaint by Zhu was the only
significant problem reported from the first ca. 500 downloads. There was
however a problem in registering for the programs via the homepage using
iPhones, but I think that I have managed to fix that now (I don't
possess an iPhone or any other cell phone). However I would be very
grateful to hear if anyone encounters difficulties with the new release
or the homepage, there is so much new material in the homepage that
there must be plenty of typos.
George
On 02/27/2013 11:21 PM, James Stroud wrote:
The underlying issue is that the implementation of OS X dynamic linking is
dependent system architecture. As I understand it, OS X 10.5.8 has a 32 bit
kernel, so its dynamic linking is not forward-compatible with 64 bit OS X
versions, for which SHELX-2013 seems to built.
Also, to be fair, it's not as if Linux users have ever been able to migrate
from from 32 bit to 64 bit without a complete system overhaul.
So while I believe that OS X may not be the optimal OS of the future, the 32/64
bit incompatibility should not be blamed as the reason.
James
p.s. What I don't understand is why the dynamic loader was ever needed in the first
place. As a "zero dependency" program, wouldn't it make sense just to
statically compile all of the SHEL* programs at the cost of a little bigger downloads?
On Feb 27, 2013, at 3:10 AM, Tim Gruene wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
phew - and people claim, macs were user-friendly. On Linux you just
add '-static' (at least for command line tools as the shelx programs)
and make the sane assumption that the kernel is less than say 10 years
old...
Tim
On 02/26/2013 04:50 PM, Guangyu Zhu wrote:
I just downloaded it. I'm using Mac 10.5.8 and getting error
messages:
dyld: unknown required load command 0x80000022 Trace/BPT trap
Google search shows that:
You should contact the developer of this application. Only the
developer can fix this. The application was incorrectly built on a
OS X 10.6 machine for a OS X 10.5 machine. The developer can fix
this by considering three things:
1. Using the correct compiler parameters: gcc-4.2
-mmacosx-version-min=10.5 -isysroot /Developer/SDKs/MacOSX10.5.sdk
...
2. Using the correct linker settings (setting environment variable
before link command). This is required, so that the OS X 10.6
linker will not use the loader command 'LC_DYLD_INFO_ONLY'
(=0x80000022), because OS X 10.5 does not understand this command:
export MACOSX_DEPLOYMENT_TARGET=10.5 (or setenv
MACOSX_DEPLOYMENT_TARGET=10.5)
After this is fixed, one can check if the application was correctly
built for OS X 10.5 by running 'otool':
otool -l binary
The correct binary should not contain any 'LC_DYLD_INFO_ONLY'
load commands (only 'LC_DYLD_INFO' commands).
(also see my blog article http://grauonline.de/wordpress/?p=71 )
Is it possible to build one for Mac 10.5.8?
Thanks!
Guangyu Zhu
On 2/26/13 3:32 AM, "George Sheldrick"
<gshe...@shelx.uni-ac.gwdg.de> wrote:
As some of you have already discovered, there is a major new
release of the whole of SHELX (the first since 1997) complete
with a new homepage that should make downloads easier. To obtain
the programs, please point your browser to:
http://shelx.uni-ac.gwdg.de/SHELX/
and then 'register' (top of blue menu, upper left). You will
then receive the password immediately by email and can then go to
'downloads'. The procedure has been designed to make life
difficult for spammers etc. A little program (in FORTRAN of
course) turns the registrations into a sorted users' list, which
I hand-edit where necessary before it appears on the homepage,
this may take a few days. The homepage also provides access to
extensive documentation, FAQs etc. Please let me know of any
problems (even typos) with the homepage and programs, with so
much new material there are sure to be some bugs. The new
versions replace all previous versions including beta-tests and
almost all the programs have been improved since their last
beta-test version (see 'recent changes').
In particular, the new shelxe_2013/2 corrects a couple of serious
bugs in the autotracing from MR models and MRSAD present in the
beta-test 2013/1.
George
-- Prof. George M. Sheldrick FRS Dept. Structural Chemistry,
University of Goettingen, Tammannstr. 4, D37077 Goettingen,
Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582
- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen
GPG Key ID = A46BEE1A
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iD8DBQFRLduMUxlJ7aRr7hoRAkNUAJ99nANlYCXYmlOqTpvKnDJyg+A1bwCdF7GA
Ag8n9o/IzBAkuX+HBfH8cMw=
=WBV3
-----END PGP SIGNATURE-----
--
Prof. George M. Sheldrick FRS
Dept. Structural Chemistry,
University of Goettingen,
Tammannstr. 4,
D37077 Goettingen, Germany
Tel. +49-551-39-3021 or -3068
Fax. +49-551-39-22582
