Hello,
I have errors in compiling einsteintoolkit.th on a mac: First, it did not configure HDF5 Compiling einsteintoolkit CST error 1: -> Configuration script for thorn HDF5 returned exit code 1 Error message: 'Error while building HDF5. Aborting.'
With external HDF5 directory, it complained Use external hdf5 --error libsz.dylib needed
With external HDF5_DIR and LIBSZ_DIR the error is:
Preprocessing /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc Compiling /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc: In function 'bool CarpetIOBasic::UseScientificNotation(const T&) [with T = int]': /Users/babiuc/Cactus/arrangements/Carpet/Carpet/src/typecase:131:1: instantiated from here /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc:701:5: internal compiler error: in tsubst_copy_and_build, at cp/pt.c:13326 Please submit a full bug report, with preprocessed source if appropriate. See http://gcc.gnu.org/bugs.html for instructions. make[3]: *** [iobasic.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/Users/babiuc/Cactus/configs/einsteintoolkit/lib/libthorn_CarpetIOBasic.a] Error 2 make: *** [einsteintoolkit] Error 2
please help, Maria
PS: I am using the new gnu compilers: CC = gcc-mp-4.6 CXX = g++-mp-4.6 F77 = gfortran-mp-4.6 F90 = gfortran-mp-4.6 and the hdf5 is the one used by the visit2.2.0 binaries for mac x86_64: hdf5-1.8.4-x86_64 and required by CarpetHDF5.
Maria
Your error message indicates that your compiler has a problem. You could install a different version of gcc, e.g. gcc 4.5.2 either from scratch or via MacPorts, and then use this compiler, which should work (I'm using it). You can have several versions of gcc installed at the same time.
-erik
On Mon, Feb 7, 2011 at 12:19 PM, Babiuc-Hamilton, Maria babiuc@marshall.edu wrote:
Hello,
I have errors in compiling einsteintoolkit.th on a mac: First, it did not configure HDF5 Compiling einsteintoolkit CST error 1: -> Configuration script for thorn HDF5 returned exit code 1 Error message: 'Error while building HDF5. Aborting.'
With external HDF5 directory, it complained Use external hdf5 --error libsz.dylib needed
With external HDF5_DIR and LIBSZ_DIR the error is:
Preprocessing /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc Compiling /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc: In function 'bool CarpetIOBasic::UseScientificNotation(const T&) [with T = int]': /Users/babiuc/Cactus/arrangements/Carpet/Carpet/src/typecase:131:1: instantiated from here /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc:701:5: internal compiler error: in tsubst_copy_and_build, at cp/pt.c:13326 Please submit a full bug report, with preprocessed source if appropriate. See http://gcc.gnu.org/bugs.html for instructions. make[3]: *** [iobasic.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/Users/babiuc/Cactus/configs/einsteintoolkit/lib/libthorn_CarpetIOBasic.a] Error 2 make: *** [einsteintoolkit] Error 2
please help, Maria
PS: I am using the new gnu compilers: CC = gcc-mp-4.6 CXX = g++-mp-4.6 F77 = gfortran-mp-4.6 F90 = gfortran-mp-4.6 and the hdf5 is the one used by the visit2.2.0 binaries for mac x86_64: hdf5-1.8.4-x86_64 and required by CarpetHDF5.
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On 7 Feb 2011, at 18:19, Babiuc-Hamilton, Maria wrote:
Hello,
I have errors in compiling einsteintoolkit.th on a mac: First, it did not configure HDF5 Compiling einsteintoolkit CST error 1: -> Configuration script for thorn HDF5 returned exit code 1 Error message: 'Error while building HDF5. Aborting.'
With external HDF5 directory, it complained Use external hdf5 --error libsz.dylib needed
With external HDF5_DIR and LIBSZ_DIR the error is:
Hi Maria,
I'm not sure I follow. Did you also try to get HDF5 to build as part of Cactus, before trying the external directories? If you omit the HDF5_DIR variable, and don't say HDF5=yes, then ExternalLibraries/HDF5 should attempt to build HDF5. Do you get an error when it tries to build this? If so, what is that error? Do you have an up-to-date version of ExternalLibraries/HDF5?
Preprocessing /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc Compiling /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc: In function 'bool CarpetIOBasic::UseScientificNotation(const T&) [with T = int]': /Users/babiuc/Cactus/arrangements/Carpet/Carpet/src/typecase:131:1: instantiated from here /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc:701:5: internal compiler error: in tsubst_copy_and_build, at cp/pt.c:13326 Please submit a full bug report, with preprocessed source if appropriate.
As Erik said, this indicates a bug in the compiler, which you could report to the GNU team.
Erik: should we consider blacklisting this compiler for the ET until we find a workaround?
See http://gcc.gnu.org/bugs.html for instructions. make[3]: *** [iobasic.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/Users/babiuc/Cactus/configs/einsteintoolkit/lib/libthorn_CarpetIOBasic.a] Error 2 make: *** [einsteintoolkit] Error 2
please help, Maria
PS: I am using the new gnu compilers: CC = gcc-mp-4.6 CXX = g++-mp-4.6 F77 = gfortran-mp-4.6 F90 = gfortran-mp-4.6 and the hdf5 is the one used by the visit2.2.0 binaries for mac x86_64: hdf5-1.8.4-x86_64 and required by CarpetHDF5.
Ian, yes I did, and obtained the error I included first in the message: Configuration script for thorn HDF5 returned exit code 1. But this problem is now solved. Thank you, Erik, the 4.6 compiler was the cause. The gcc 4.5 works just fine.
thanks, Maria
________________________________________ From: Ian Hinder [ian.hinder@aei.mpg.de] Sent: Tuesday, February 08, 2011 2:47 AM To: Babiuc-Hamilton, Maria Cc: users@einsteintoolkit.org Subject: Re: [Users] Error in compiling einsteintoolkit on Mac
On 7 Feb 2011, at 18:19, Babiuc-Hamilton, Maria wrote:
Hello,
I have errors in compiling einsteintoolkit.th on a mac: First, it did not configure HDF5 Compiling einsteintoolkit CST error 1: -> Configuration script for thorn HDF5 returned exit code 1 Error message: 'Error while building HDF5. Aborting.'
With external HDF5 directory, it complained Use external hdf5 --error libsz.dylib needed
With external HDF5_DIR and LIBSZ_DIR the error is:
Hi Maria,
I'm not sure I follow. Did you also try to get HDF5 to build as part of Cactus, before trying the external directories? If you omit the HDF5_DIR variable, and don't say HDF5=yes, then ExternalLibraries/HDF5 should attempt to build HDF5. Do you get an error when it tries to build this? If so, what is that error? Do you have an up-to-date version of ExternalLibraries/HDF5?
Preprocessing /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc Compiling /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc: In function 'bool CarpetIOBasic::UseScientificNotation(const T&) [with T = int]': /Users/babiuc/Cactus/arrangements/Carpet/Carpet/src/typecase:131:1: instantiated from here /Users/babiuc/Cactus/arrangements/Carpet/CarpetIOBasic/src/iobasic.cc:701:5: internal compiler error: in tsubst_copy_and_build, at cp/pt.c:13326 Please submit a full bug report, with preprocessed source if appropriate.
As Erik said, this indicates a bug in the compiler, which you could report to the GNU team.
Erik: should we consider blacklisting this compiler for the ET until we find a workaround?
See http://gcc.gnu.org/bugs.html for instructions. make[3]: *** [iobasic.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/Users/babiuc/Cactus/configs/einsteintoolkit/lib/libthorn_CarpetIOBasic.a] Error 2 make: *** [einsteintoolkit] Error 2
please help, Maria
PS: I am using the new gnu compilers: CC = gcc-mp-4.6 CXX = g++-mp-4.6 F77 = gfortran-mp-4.6 F90 = gfortran-mp-4.6 and the hdf5 is the one used by the visit2.2.0 binaries for mac x86_64: hdf5-1.8.4-x86_64 and required by CarpetHDF5.
-- Ian Hinder ian.hinder@aei.mpg.de
On Tue, Feb 8, 2011 at 2:47 AM, Ian Hinder ian.hinder@aei.mpg.de wrote:
On 7 Feb 2011, at 18:19, Babiuc-Hamilton, Maria wrote:
Hello,
I have errors in compiling einsteintoolkit.th on a mac: First, it did not configure HDF5 Compiling einsteintoolkit CST error 1: -> Configuration script for thorn HDF5 returned exit code 1 Error message: 'Error while building HDF5. Aborting.'
With external HDF5 directory, it complained Use external hdf5 --error libsz.dylib needed
With external HDF5_DIR and LIBSZ_DIR the error is:
Hi Maria,
I'm not sure I follow. Did you also try to get HDF5 to build as part of Cactus, before trying the external directories? If you omit the HDF5_DIR variable, and don't say HDF5=yes, then ExternalLibraries/HDF5 should attempt to build HDF5. Do you get an error when it tries to build this? If so, what is that error? Do you have an up-to-date version of ExternalLibraries/HDF5?
By default, ExternalLibraries/HDF5 looks for an existing HDF5 installation and performs a few basic tests to see whether this installation works. It does not check whether compiler versions etc. match, so it is possible that it finds an incompatible HDF5. To ensure that the HDF5 library is built, you have to say HDF5_DIR=BUILD.
Should we change this? The only disadvantage of building HDF5 is that this takes time.
-erik
users@lists.einsteintoolkit.org