dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

On 02/02/11 10:55, Anders Logg wrote:
> On Mon, Jan 31, 2011 at 09:40:19AM +0100, Anders Logg wrote:
>> I'm working on a fix. The problem wasn't visible when DOLFIN_NOPLOT
>> was not set.
> 
> This should now be fixed. There might still be some minor issues that
> the buildbot will pick up.
> 
> Here's a short summary of what has changed:
> 
> 1. The refine() function returns a reference to the refined mesh
> instead of return-by-value in C++. The following no longer works:
> 
>   mesh = refine(mesh);
> 
> Instead, one must do
> 
>   Mesh& refined_mesh = refine(mesh);
> 
> or equivalently
> 
>   refine(mesh);
>   Mesh& refined_mesh = mesh.child();
> 

Can get a shared pointer? Otherwise we're in trouble if mesh goes out of
scope.

Garth

> 2. The refine() function works as expected in Python:
> 
>   mesh = refine(mesh)
> 
> should work fine since refine() returns a shared_ptr in Python.
> 
> 3. A hierarchy of meshes is now stored as part of the mesh, accessed
> by the parent() and child() functions. There are also coarse() and
> fine() functions for accessing the coarsest and finest meshes in a
> hierarchy.
> 
> 4. Other classes like Function, FunctionSpace, Form, VariationalProblem
> can now also been refined. This functionality is untested and will
> need some further debugging before it can be used in the
> AdaptiveVariationalSolver class.
> 
> --
> Anders
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~dolfin
> Post to     : dolfin@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~dolfin
> More help   : https://help.launchpad.net/ListHelp