FYI, the issue appears to be either fixed or latent in current clang/llvm trunk for the 7.0 release. However Apple is conservative about absorbing changes from llvm.org so we need to file a radar to make sure they test that this issue if fixed in the next major Xcode release.
FYI, the preprocessed failing test case doesn't trigger on ubuntu bionic using the same clang 6.0.0 release so the issue appears to be entirely darwin specific. The proposed fix of restricting the addition of -fno-strict-aliasing to darwin should be appropriate.
Note that I opened https://bugs.llvm.org/show_bug.cgi?id=37280 with the upstream llvm clang developers.
I would also note that, on darwin, we seem to automatically get -fno-strict-aliasing appended to compilations of c code through setup.py but not c++ code. Hence the need to explicitly pass -fno-strict-aliasing in setup.py on darwin. /usr/bin/clang -fno-strict-aliasing -fno-common -dynamic -pipe -Os -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -D_PYMOL_LIBPNG -D_PYMOL_INLINE -D_PYMOL_VMD_PLUGINS -D_HAVE_LIBXML -D_PYMOL_FREETYPE -DNO_MMLIBS -D_PYMOL_NUMPY -D_PYMOL_OSX -Iov/src -Ilayer0 -Ilayer1...
I would also note that, on darwin, we seem to automatically get -fno-strict-aliasing appended to compilations of c code through setup.py but not c++ code. Hence the need to explicitly pass -fno-strict-aliasing in setup.py on darwin. /usr/bin/clang -fno-strict-aliasing -fno-common -dynamic -pipe -Os -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -D_PYMOL_LIBPNG -D_PYMOL_INLINE -D_PYMOL_VMD_PLUGINS -D_HAVE_LIBXML -D_PYMOL_FREETYPE -DNO_MMLIBS -D_PYMOL_NUMPY -D_PYMOL_OSX -Iov/src -Ilayer0 -Ilayer1...
This issue can be fixed far more gracefully with the following change to setup.py... Index: setup.py =================================================================== --- setup.py (revision 4187) +++ setup.py (working copy) @@ -392,12 +392,12 @@ libs += ["GLEW"] libs += pyogl_libs - ext_comp_args += ["-ffast-math", "-funroll-loops", "-fcommon"] + ext_comp_args += ["-ffast-math", "-funroll-loops", "-fcommon", "-O3"] # optimization currently causes a clang segfault on OS X 10.9 when # compiling layer2/RepCylBond.cpp...
Opps. I missed that you are disabling -O3 on darwin. It would be better to enhance that with a check the offending compiler version of clang instead of all versions of clang.
modules not built with -O3 like main pymol files
The problem is in the fink and macports packaging which was based on a much earlier behavior of PYMOL_PATH. Setting it to PYMOL_PATH=/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pymol/pymol_path instead of PYMOL_PATH=/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pymol eliminated the need for the redundant installation of the subdirectories associated with pymol_path.
Does the incentive build install the modules subdirectory? I don't recall it having APBS Tools plugin. Perhaps it has been omitted by accident because of that.
This is using the MacPorts python packaging for the 2.7 and 3.x series using the Python port group for installing python packages. https://guide.macports.org/chunked/reference.portgroup.html which should be installing pymol with.... ${python.bin} setup.py install --no-user-cfg --prefix=${python.prefix} --root=${destroot} Note that the fink packaging system sees the same issue... https://github.com/fink/fink-distributions/blob/master/10.9-libcxx/stable/main/finkinfo/sci/pymol-py.info and the modules...
This is using the MacPorts python packaging for the 2.7 and 3.x series using the Python port group for installing python packages. https://guide.macports.org/chunked/reference.portgroup.html which should be installing pymol with.... ${python.bin} setup.py --no-user-cfg --prefix=${python.prefix} --root=${destroot}
The actual problem stems from setup.py not installing the modules directory. Is there a reason why we have to manually install that directory?
FYI, the current 2.1.0 release with PyQt support can be built under MacPorts with the proposed packaging changes at https://trac.macports.org/ticket/56139 .
FYI, the current 2.1.0 release with PyQt support can be built under MacPorts with the proposed packaging changes at https://trac.macports.org/ticket/56139.
pymol install doesn't compile python scripts in modules
I have managed to get the APBS plugin to work partially under a MacPorts build of pymol with the PyQT interface as long the tk and pmg components are installed. However,, while I can use the APBS plugin to generate electrostatic maps for a pdb file, when I try to display the electrostatic surface, pymol goes into an endless cycle of progress bars being displayed. Have you been able to get the APBS plugin to render surfaces when pymol is run under the PyQt interface?
I have managed to get the APBS plugin to work partially under a MacPorts build of pymol with the PyQT interface as long the tk and pmg components are installed. However,, while I can use the APBS plugin to generate electrostatic maps for a pdb file, when I try to display the electrostatic surface, pymol goes into an endless cycle . of progress bars being displayed. Have you been able to get the APBS plugin to render surfaces when pymol is run under the PyQt interface?
Just to clarify the situation with building pymol with PyQt support on MacPorts vs fink, currently only MacPorts will be able to produce a working pymol using the PyQt support. The issue appears to be that fink's glew package is built against the X11 libGL (which is a wrappered version providing additional X11 support to libGL) whereas the MacPorts glew package is built against the system libGL instead. Also, it is necessary to build pymol with the --osx-framworks option passed to setup.py to avoid...
Note that the MacPorts test packaging for pymol 2.1.0 at https://trac.macports.org/ticket/56139 reveals that when built for X11 tk support, the resulting pymol, when started in the presence of PyQt, will fail to properly initialize its GL support.... Detected OpenGL version 2.0 or greater. Shaders available. Detected GLSL version 0.0. OpenGL graphics engine: GL_VENDOR: (null) GL_RENDERER: (null) GL_VERSION: (null) Detected 8 CPU cores. Enabled multithreaded rendering. Deinstalling the PyQt package...
Note that I have posted initial packaging changes to the MacPorts pymol package for builds with either the X11 or PyQt support. https://trac.macports.org/ticket/56139
run time option to disable use of PyQt should be added
QT support not fully decoupled from Tk support
This will be useful to have in fink for building the upcoming gcc8-8.1.0-1 packaging.
Info file for libmpfr5.info
add libmpfr5/libmpfr5-shlibs packaging for mpfr 4.0.0
Revised as requested Index: scipy-py.info =================================================================== RCS file: /cvsroot/fink/dists/10.9-libcxx/stable/main/finkinfo/sci/scipy-py.info,v retrieving revision 1.12 diff -u -r1.12 scipy-py.info --- scipy-py.info 23 Jun 2017 20:34:52 -0000 1.12 +++ scipy-py.info 30 Nov 2017 02:38:17 -0000 @@ -1,6 +1,6 @@ Info2: << Package: scipy-py%type_pkg[python] -Version: 0.19.1 +Version: 1.0.0 Epoch: 1 Revision: 1 Maintainer: None <fink-devel@lists.sourceforge.net>...
cmake-3.10.0-1
Revised patch file for ccpnmr-py-2.4.2-3 to switch build to gcc6
Info file for ccpnmr-py-2.4.2-3 to switch build to gcc6
ccpnmr-py-2.4.2-3 to switch build to gcc6
Info file for cython-py-0.27.3-1
cython-py-0.27.3-1
Hanspeter, if you are going to suppress the installation of the pyc files during the InstallScript, you really should have added PostInstScript to generate them on package installation and a PreRmScript to delete them on package removal. Please fix that damage to the packaging.
Info file for scipy-py-1.0.0-1 update fixed for fink-bld.
Use fix proposed by Derek Homeier to manually set SHELL=/bin/sh for compilation. Index: scipy-py.info =================================================================== RCS file: /cvsroot/fink/dists/10.9-libcxx/stable/main/finkinfo/sci/scipy-py.info,v retrieving revision 1.12 diff -u -r1.12 scipy-py.info --- scipy-py.info 23 Jun 2017 20:34:52 -0000 1.12 +++ scipy-py.info 6 Nov 2017 12:18:23 -0000 @@ -1,6 +1,6 @@ Info2: << Package: scipy-py%type_pkg[python] -Version: 0.19.1 +Version: 1.0.0 Epoch:...
Info file for numpy-py27-1.13.3-1 fixed for fink-bld
Use fix proposed by Derek Homeier to manually set SHELL=/bin/sh for compilation. Index: numpy-py.info =================================================================== RCS file: /cvsroot/fink/dists/10.9-libcxx/stable/main/finkinfo/sci/numpy-py.info,v retrieving revision 1.12 diff -u -r1.12 numpy-py.info --- numpy-py.info 19 Mar 2017 21:49:31 -0000 1.12 +++ numpy-py.info 6 Nov 2017 12:14:18 -0000 @@ -1,6 +1,6 @@ Info2: << Package: numpy-py%type_pkg[python] -Version: 1.12.1 +Version: 1.13.3 Revision:...
wget-1.19.2-1
scipy-py-1.0.0-1 update
update pytest-py to 3.2.3 release for scipy 1.0.0
add attrs-py package to allow pytest-py update
hypothesis-py-3.33.0-1 update for pytest-py update
py-py-1.4.34-1 update for newer pytest
telnet-54.50.1-2 to build skinny binaries
On Wed, Oct 25, 2017 at 7:37 AM, Daniel Macks dmacks@users.sf.net wrote: Unrelatedly, the build gives a fat binary. Any way we can scrap the i386 to save a bunch of compile-time? Daniel, I assume this could be done by changing the line in libtelnet/libtelnet-13/libtelnet.xcodeproj/project.pbxproj from... ARCHS = "$(ARCHS_STANDARD_32_64_BIT)"; to ARCHS = "$(ARCHS_STANDARD)"; per https://pewpewthespells.com/blog/buildsettings.html. Jack [bugs:#1659] telnet-54.50.1-1 Status: closed-accepted Group: Added_to_Fink...
I can reproduce this issue if I have the fink 'make' package installed so the trivial fix for now is just to add... Index: telnet.info =================================================================== RCS file: /cvsroot/fink/dists/10.9-libcxx/stable/main/finkinfo/net/telnet.info,v retrieving revision 1.1 diff -u -r1.1 telnet.info --- telnet.info 10 Oct 2017 17:25:03 -0000 1.1 +++ telnet.info 25 Oct 2017 11:05:30 -0000 @@ -21,7 +21,7 @@ cd ../../libtelnet-13 xcodebuild -project libtelnet.xcodeproj...
Doesn't this imply that the ray-trace demo needs to be tweaked to provide a more uniform experience? Having a demo which displays a checker board background when run but a black one when clicked on, seems a bit random in behavior.
Info file for lzip-1.19-1
lzip-1.19-1
Info file for gawk-4.2.0-1
gawk-4.2.0-1
pymol 1.9.0.0 produces checkerboard in ray-tracing demo in HD 2600 XT
dejagnu-1.6.1-1
Info file for gcc5-5.5.0-1
gcc5-5.5.0-1