Re: [moose-generic] MOOSE_installation_problem

[Subhasis R. <ray...@gm...>]

Dear Vadim,

 On 3/28/13, Vadim A <ast...@nc...> wrote:
> 1. I modified the code
>
>     /// Go through all elements under /classes and ask for defining a
>     /// Python class for it.
>     static int defineAllClasses(PyObject * module_dict)
>     {
>         ///static vector <Id> classes(Field< vector<Id>
>>::get(ObjId("/classes"),"children"));
>         static const string children("children");
>         static const string path("/classes");
>         static const ObjId cle(path);
>         assert(!(cle == ObjId::bad()));
>         static vector <Id> classes = Field<vector<Id> >::get(cle,children);
>         //static vector <Id> classes = Field<vector<Id>
>>::get(ObjId(path),children);
>
>
>
> What should I expect to see with that assert ? I did not noticed any
> differences when I did compile and import.
>
>

I was suspecting a static initialization order issue. Had the
"/classes" element not been created by the time defineAllClasses tries
to access it, the program would have exited with an assertion failure
in stead of segmentation fault. But  from your results it does not
seem to be the problem.

>
> Also, when I do inspect core , I do see some references to Intel compiler,
> can it be indication of some problem ?
>

Since gdb is loading libraries related to intel compiler before
python, is it possible that you built  Python with intel compiler? In
that case can you see if rebuilding it with gcc and then building
moose helps?

Best,
  Subha