Re: [Libmesh-devel] Is EquationSystems::delete_system() broken?

[John P. <pet...@cf...>]

Roy Stogner writes:
 > On Mon, 22 Jan 2007, John Peterson wrote:
 > 
 > > I think it would have the same number if it had the same name or a
 > > later one in the alphabet.  I think they are stored alphabetically in
 > > a std::map.  So
 > >
 > > es.add_system("B");
 > > es.add_system("C");
 > >
 > > B = system 0
 > > C = system 1
 > >
 > > es.delete_system("C");
 > > es.add_system("A");
 > >
 > > A=system 0
 > > C=system 1
 > 
 > Are you sure?  EquationSystems::add_system just seems to pick
 > this->n_systems() as the index to use when constructing a new System
 > object.

Oh yeah.  The number of the system is set via the System constructor,
not dynamically via its map ordering.

 > Plus, after adding a system to the EquationSystems, it gets added via
 > _add_system_to_nodes_and_elems() to each DofObject in the mesh - but
 > none of the other systems seem to get touched, as one would expect to
 > happen if they might have been renumbered.
 > ---
 > Roy