[pooma-dev] Example of bad behavior from particle boundary conditions

[Julian C. Cummings]

Hmmm... I took your modified version of
Bounce.cpp from the attachment and built
it under the Intel VTune compiler.  It
seems to compile and run just fine.

Unfortunately, I do not have a copy of
CW Pro 6.1 for the PC to test with.  Are
you able to run the CW debugger and see
what is going wrong?

-- Julian C.


-----Original Message-----
From: Steve Nolen [mailto:drnuke at lanl.gov]
Sent: Wednesday, April 04, 2001 12:53 PM
To: Pooma
Subject: [pooma-dev] Example of bad behavior from particle boundary
conditions


I have made a small adjustment to 'Bounce.cpp' to show the problem i'm
seeing.  i am compiling under Metrowerks 6.1 for the PC.  if I move the new
call, addBCs, around it exhibits different behavior.

i believe the problem may originate from a call in ParticleBC.h shown below.

  // Factory method to create the boundary condition.
  template <class Subject, class Object>
  ParticleBCItem* create(const Subject& s, const Object& o) const
  {
    typedef ParticleBC<Subject,Object,BCType> PBC_t;
    return new PBC_t(s, o, static_cast<const BCType&>(*this));
  }

the pointer is stored in a stl vector of bc pointers within the particle
base class.

something about this doesn't look quite right.

-steve