Ola Skavhaug wrote: > LUK ShunTim skrev den 16/01-2008 følgende: [snipped]
>> For dolfin, the c++ library builds all right but pydolfin gives this error >> >> "/usr/bin/swig -c++ -python -ignoremissing -nodefaultctor -nodefaultdtor >> -dirprot -Werror `pkg-config --cflags ufc-1` -I/usr/include >> -I../../src/kernel/common -I../../src/kernel/elements >> -I../../src/kernel/fem -I../../src/kernel/function >> -I../../src/kernel/graph -I../../src/kernel/io -I../../src/kernel/la >> -I../../src/kernel/main -I../../src/kernel/math -I../../src/kernel/mesh >> -I../../src/kernel/mf -I../../src/kernel/nls -I../../src/kernel/ode >> -I../../src/kernel/pde -I../../src/kernel/quadrature >> -I../../src/kernel/parameter -I../../src/kernel/log >> -I../../src/kernel/plot -DHAVE_SCOTCH_H=1 -DHAVE_MPI_H=1 >> -DMPICH_IGNORE_CXX_SEEK -DHAVE_PETSC_H=1 -DNDEBUG -o dolfin_wrap.cpp >> dolfin.i >> ../../src/kernel/la/dolfin/Scalar.h:88: Warning(503): Can't wrap >> 'operator real' unless renamed to a valid identifier. >> make[2]: *** [dolfin_wrap.cpp] Error 1" >> >> Is this a bug? > > I get this as well, but Anders, for instance, does not. The simple fix is to > type make once more, as all warnings are treated as errors, but the swig > generated wrapper code will not be regenerated at the second make (and hence > the warning vanishes). > > And yes, I consider this a bug. Hello Ola, Thanks. I'm not familiar with swig but your hint guide me to read the man page and the error messages carefully again. Indeed the -Werror option *is passed* to swig, making all warnings fatal. While the second make took me past this error, further down the road it failed again with "cc1plus: warnings being treated as errors". I think the culprit is the -Werror option in these 2 lines "CXXFLAGS = @DOLFIN_DEFINITIONS@ -g -O2 -fno-strict-aliasing -Wall -Werror -ansi -Wno-uninitialized" " $(SWIG) $(SWIG_PYTHON_OPT) -ignoremissing -nodefaultctor -nodefaultdtor -dirprot -Werror $(UFC_PATH) @KERNEL_INCLUDE@ @DOLFIN_DEFINITIONS@ -o $@ $<" in pydolfin/Makefile.in. Thanks again and regards, ST -- _______________________________________________ DOLFIN-dev mailing list [email protected] http://www.fenics.org/mailman/listinfo/dolfin-dev
