compiling problem with latest cctbx on OS X
With Ralf and Erik's help and advice, (cf http://www.phenix-
online.org/pipermail/cctbxbb/2005-September/thread.html ) I made a
Mac OS X fink package for the cctbx source release.
http://pdb.finkproject.org/pdb/search.php?summary=cctbx
Everything works fine on OS X v. 10.4, gcc 4.0.1, on both ppc and
intel platforms, until today when I realized the package I made is
now out of date (as of July I think).
I'm now having problems compiling the latest version. I've put the
log file here:
http://xanana.ucsc.edu/~wgscott/kipaut/fink/compile.log
Briefly, I think the failure is taking place here:
c++ -o cctbx/translation_search/boost_python/fast_nv1995.o -c -fPIC -
no-cpp-precomp -ftemplate-depth-130 -Wno-long-double -w -
DBOOST_DISABLE_THREADS -DNDEBUG -O3 -ffast-math -
DBOOST_PYTHON_MAX_BASES=2 -I/sw/src/fink.build/cctbx-60705-1001/
cctbx_sources/boost -Iinclude -I/sw/src/fink.build/cctbx-60705-1001/
cctbx_sources/include -I/sw/src/fink.build/cctbx-60705-1001/
cctbx_sources/cctbx/include -I/sw/src/fink.build/cctbx-60705-1001/
cctbx_sources/scitbx/include -I/sw/src/fink.build/cctbx-60705-1001/
cctbx_sources -I/Library/Frameworks/Python.framework/Versions/2.4/
include/python2.4 /sw/src/fink.build/cctbx-60705-1001/cctbx_sources/
cctbx/translation_search/boost_python/fast_nv1995.cpp
/sw/src/fink.build/cctbx-60705-1001/cctbx_sources/scitbx/include/
scitbx/fftpack/complex_to_complex.h: In member function 'void
scitbx::fftpack::complex_to_complex
--- William Scott
Briefly, I think the failure is taking place here:
c++ -o cctbx/translation_search/boost_python/fast_nv1995.o -c -fPIC - no-cpp-precomp -ftemplate-depth-130 -Wno-long-double -w - DBOOST_DISABLE_THREADS -DNDEBUG -O3 -ffast-math - DBOOST_PYTHON_MAX_BASES=2 -I/sw/src/fink.build/cctbx-60705-1001/ cctbx_sources/boost -Iinclude -I/sw/src/fink.build/cctbx-60705-1001/ cctbx_sources/include -I/sw/src/fink.build/cctbx-60705-1001/ cctbx_sources/cctbx/include -I/sw/src/fink.build/cctbx-60705-1001/ cctbx_sources/scitbx/include -I/sw/src/fink.build/cctbx-60705-1001/ cctbx_sources -I/Library/Frameworks/Python.framework/Versions/2.4/ include/python2.4 /sw/src/fink.build/cctbx-60705-1001/cctbx_sources/ cctbx/translation_search/boost_python/fast_nv1995.cpp /sw/src/fink.build/cctbx-60705-1001/cctbx_sources/scitbx/include/ scitbx/fftpack/complex_to_complex.h: In member function 'void scitbx::fftpack::complex_to_complex
::pass3 (scitbx::fftpack::select_sign<Tag>, size_t, size_t, RealType*, RealType*, const RealType*, const RealType*)': /sw/src/fink.build/cctbx-60705-1001/cctbx_sources/scitbx/include/ scitbx/fftpack/complex_to_complex.h:361: error: expected unqualified- id before numeric constant
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002. Could you please look at the output of "limit"? The default stacksize used to be too small under 10.2 for sure, and 10.3 maybe. I don't know about 10.4. Here is what I use to correct the problem: limit datasize unlimited limit stacksize 8192k If that doesn't help, what exactly is your environment? 10.4? G5 or Intel? How did you install the developer tools? Thanks! Cheers, Ralf __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On Aug 29, 2006, at 2:30 PM, Ralf W. Grosse-Kunstleve wrote:
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002.
Could you please look at the output of "limit"? The default stacksize used to be too small under 10.2 for sure, and 10.3 maybe. I don't know about 10.4. Here is what I use to correct the problem:
limit datasize unlimited limit stacksize 8192k
If that doesn't help, what exactly is your environment? 10.4? G5 or Intel? How did you install the developer tools?
Thanks!
Cheers, Ralf
Hi Ralf: That might be part of the problem. Last time I built it under (ba) sh, which doesn't appear to have a limit command, unlike tcsh and zsh. So I am rebuilding under tcsh now (I had encountered a problem with zsh previously but I can't recall what it was). For the record, this is the latest Developer Tools, version 2.4, standard default install. Same error appears on both platforms (ppc and intel, both are G5's). gcc --version powerpc-apple-darwin8-gcc-4.0.1 (GCC) 4.0.1 (Apple Computer, Inc. build 5363) g++ --version powerpc-apple-darwin8-g++-4.0.1 (GCC) 4.0.1 (Apple Computer, Inc. build 5363) % sw_vers ProductName: Mac OS X ProductVersion: 10.4.7 BuildVersion: 8J135 Right now I am redoing this with tcsh and the limit command you suggested. Hopefully that will cure the problem (which didn't appear for any previous version, btw). Bill
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002.
Could you please look at the output of "limit"?
same problem under tcsh I am afraid. For the record, I now have % limit cputime unlimited filesize unlimited datasize unlimited stacksize 8192 kbytes coredumpsize 0 kbytes memoryuse unlimited descriptors 256 memorylocked unlimited maxproc 100
On Aug 29, 2006, at 2:30 PM, Ralf W. Grosse-Kunstleve wrote:
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002.
As a positive control, I just recompiled the version dated 2005.12.29 without the error occurring on the same machine and under the same conditions under which the new version has this failure.
--- William Scott
On Aug 29, 2006, at 2:30 PM, Ralf W. Grosse-Kunstleve wrote:
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002.
As a positive control, I just recompiled the version dated 2005.12.29 without the error occurring on the same machine and under the same conditions under which the new version has this failure.
Sorry, I have no clue why the compiler is choking. We use gcc 4.0.0 under Linux all the time. I think this must be a compiler bug introduced in the Apple version (not the first time). We have to figure out a workaround. In the next couple of weeks we should get a new Macintel with 10.4. I'll try to find a workaround once we have the machine. Thanks for running all the tests! Ralf __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On Aug 29, 2006, at 8:43 PM, Ralf W. Grosse-Kunstleve wrote:
--- William Scott
wrote: On Aug 29, 2006, at 2:30 PM, Ralf W. Grosse-Kunstleve wrote:
This error doesn't make sense. The particular source file (complex_to_complex.h) was last modified in Sep 2002.
As a positive control, I just recompiled the version dated 2005.12.29 without the error occurring on the same machine and under the same conditions under which the new version has this failure.
Sorry, I have no clue why the compiler is choking. We use gcc 4.0.0 under Linux all the time. I think this must be a compiler bug introduced in the Apple version (not the first time). We have to figure out a workaround. In the next couple of weeks we should get a new Macintel with 10.4. I'll try to find a workaround once we have the machine. Thanks for running all the tests! Ralf
It compiled find under ubuntu linux, fwiw, gcc --version gcc (GCC) 4.0.3 (Ubuntu 4.0.3-1ubuntu5) % uname -o -m -r 2.6.15-26-686 i686 GNU/Linux
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
participants (2)
-
Ralf W. Grosse-Kunstleve
-
William Scott