On Fri, 23 Mar 2012, Jungho Lee wrote: > The problem is that the compiler crashes as seen below.
In that case the solution is mostly likely going to come from the compiler developer. Even the most horribly broken source code isn't supposed to crash the compiler. > I can compile this individual compile if I use the flag -O0 but I am > perplexed as to why this is trying to use -O2 even though I > specified METHOD=dbg. For reference, my mpicc/mpicxx does not have > any optimization flag: > > $ mpicxx -show > clang++ -pipe -w -g -arch x86_64 -L/opt/local/lib > -I/opt/local/include/mpich2 -L/opt/local/lib -lmpichcxx -lpmpich > -lmpich -lopa -lmpl -lpthread I don't think this one's a libMesh bug either, because there's something sticking a *bunch* of extra non-libMesh-related flags in your compile command line right around that O2: > -fmodule-cache-path > /var/folders/r1/v4j3z5c939q3_w__0q67d7fm0000gn/T/clang-module-cache > -O2 -w -fdeprecated-macro -fdebug-compilation-dir > /Users/jungholee/projects/herd/libmesh -ferror-limit 19 > -fmessage-length 118 -stack-protector 1 -fblocks > -fobjc-runtime-has-arc -fobjc-runtime-has-weak > -fobjc-dispatch-method=mixed -fcxx-exceptions -fexceptions > -fdiagnostics-show-option -fcolor-diagnostics Something scarfed from your PETSc configuration, perhaps? --- Roy ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Libmesh-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libmesh-users
