On Thu, Nov 8, 2012 at 9:55 AM, Kirk, Benjamin (JSC-EG311) <benjamin.kirk-1@nasa.gov> wrote:
Any chance you can manually hack -g in there?

./configure libmesh_CXXFLAGS="-g"

will do it I think...

Sure, will try that next, although I claim the stack trace will be no more illuminating ;-)

And in case anyone's curious, here are the different link lines used in the automake and trunk checkouts (with some stuff removed for readability)

/bin/sh ./libtool \
  --tag=CXX \
  --mode=link \
  mpicxx \
  -O2 \
  -felide-constructors \
  -funroll-loops \
  -fstrict-aliasing \
  -std=c++0x \
  -Wdisabled-optimization \
  -fopenmp \
  -o libmesh.la \
  (various .lo objects) \
  (various .la objects) \
  -lz \
  -L/opt/packages/petsc/3.3-p4/lib -lpetsc \
  -L/opt/packages/hypre/2.8.0b/lib -lHYPRE \
  -L/opt/packages/mpich2/mpich2-1.4.1p1/lib \
  -L/opt/packages/gcc_4.6.3/lib/gcc/x86_64-apple-darwin12.0.0/4.6.3 \
  -L/opt/packages/gcc_4.6.3/lib \
  -lflapack -lfblas -lmpichf90 -lpthread -lgfortran \
  -lquadmath -lm -lmpichcxx -lstdc++ -lpmpich -lmpich \
  -lopa -lmpl -lSystem -lgcc_ext.10.5 -ldl

libtool \
  --tag=CXX \
  --quiet \
  --mode=link \
  mpicxx \
  -dynamiclib \
  -Wl,-undefined,dynamic_lookup,-flat_namespace \
  -o /Users/moose/projects/libmesh/lib/x86_64-apple-darwin12.0.0_opt/libmesh.dylib \
  (various .o files) \

So I'd say the important differences are the following:

1.) The libtool used (trunk uses my hand-built one, automake uses one that has been generated in libmesh's top-level directory)

2.) automake branch passes more system libraries on the link line ( -lSystem, -lgcc_ext.10.5, -ldl, -lgfortran) than trunk.

I guess next I'll try to have my hand-built libtool (rather than libmesh's) first in my PATH, see if that makes a difference...