Mihai,
independently of Guido's advice (and he knows better than me, since he wrote
the MeshWorker framework), could you try to come up with the smallest
possible program that reproduces the problem below? It doesn't have to
do anything useful, just trigger the assertion. I'll try to see what
can be done and put it into the testsuite.
Thanks
W.
> I compiled the code (with revision 21616) and I am now faced with this
> error:
>
>
>
> --------------------------------------------------------
> An error occurred in line <383> of file
> </Users/mihai/Utils/deal.II.svn/deal.II/deal.II/include/grid/tria_accessor.templates.h>
> in function
> dealii::InvalidAccessor<<anonymous>, <anonymous>, <anonymous>
> >::InvalidAccessor(const dealii::Triangulation<dim, spacedim>*, int,
> int, const typename dealii::TriaAccessorBase<<anonymous>, <anonymous>,
> <anonymous> >::AccessorData*) [with int structdim = 0, int dim = 1,
> int spacedim = 1]
> The violated condition was:
> false
> The name and call sequence of the exception was:
> ExcMessage ("You are attempting an illegal conversion between "
> "iterator/accessor types. The constructor you call " "only exists to
> make certain template constructs " "easier to write as dimension
> independent code but " "the conversion is not valid in the current
> context.")
> Additional Information:
> You are attempting an illegal conversion between iterator/accessor
> types. The constructor you call only exists to make certain template
> constructs easier to write as dimension independent code but the
> conversion is not valid in the current context.
>
>
> The stacktrace information is not too helpful (line numbers are not
> consistent) - probably it's got something to do with the compiler I'm
> using (debug mode is enabled)...
>
>
>
> Stacktrace:
> -----------
> #0 3 step-12-1D 0x000000000001092b
> _ZN6dealii15InvalidAccessorILi0ELi1ELi1EEC1EPKNS_13TriangulationILi1ELi1EEEiiPKv
> + 121: 3 step-12-1D 0x000000000001092b
> _ZN6dealii15InvalidAccessorILi0ELi1ELi1EEC1EPKNS_13TriangulationILi1ELi1EEEiiPKv
> #1 4 step-12-1D 0x000000000001099d
> _ZN6dealii15TriaRawIteratorINS_15InvalidAccessorILi0ELi1ELi1EEEEC2Ev +
> 51: 4 step-12-1D 0x000000000001099d
> _ZN6dealii15TriaRawIteratorINS_15InvalidAccessorILi0ELi1ELi1EEEEC2Ev
> #2 5 step-12-1D 0x00000000000109b5
> _ZN6dealii12TriaIteratorINS_15InvalidAccessorILi0ELi1ELi1EEEEC1Ev +
> 21: 5 step-12-1D 0x00000000000109b5
> _ZN6dealii12TriaIteratorINS_15InvalidAccessorILi0ELi1ELi1EEEEC1Ev
> #3 6 step-12-1D 0x00000000000109ef
> _ZN6dealii10MeshWorker7DoFInfoILi1ELi1EdEC1INS_10DoFHandlerILi1ELi1EEEEERKT_
> + 55: 6 step-12-1D 0x00000000000109ef
> _ZN6dealii10MeshWorker7DoFInfoILi1ELi1EdEC1INS_10DoFHandlerILi1ELi1EEEEERKT_
> #4 7 step-12-1D 0x000000000001f6f3
> _ZN6Step12ILi1EE15assemble_systemEv + 229: 7 step-12-1D
> 0x000000000001f6f3 _ZN6Step12ILi1EE15assemble_systemEv
> #5 8 step-12-1D 0x000000000001fa52
> _ZN6Step12ILi1EE3runEv + 264: 8 step-12-1D
> 0x000000000001fa52 _ZN6Step12ILi1EE3runEv
> #6 9 step-12-1D 0x0000000000000ff0 main +
> 38: 9 step-12-1D 0x0000000000000ff0 main
> #7 10 step-12-1D 0x0000000000000f74 start +
> 52: 10 step-12-1D 0x0000000000000f74 start
> #8 11 ??? 0x0000000000000001 0x0 +
> 1: 11 ??? 0x0000000000000001 0x0
>
>
>
>
>
> Von: Wolfgang Bangerth <[email protected]>
> An: [email protected]
> CC: mihai alexe <[email protected]>; Toby D. Young
> <[email protected]>
> Gesendet: Donnerstag, den 5. August 2010, 20:00:25 Uhr
> Betreff: Re: [deal.II] MeshWorker framework for 1D problems
>
>
> Mihai,
>
> > thanks for clearing things up. guess I have to code the integration
> logic
> > by hand (along the lines of the pre-MeshWorker version of step-12).
>
> Actually, I think you probably *should* use the MeshWorker framework.
> The
> reason you get linker errors is simply because the classes you are
> missing
> are not instantiated for 1d, see for example the file
> deal.II/source/numerics/mesh_worker_info.cc. Instead of reinventing
> the
> wheel, let's just try to these instantiations work in 1d as well.
>
> So I've made things to at least compile in revision 21614. By simply
> disabling
> all face integration stuff since the FEFaceValues and FESubFaceValues
> classes
> don't work in 1d. Now, I'm under no illusion that things will work in
> 1d
> because integration has to work differently there. Can you try out how
> far
> you get when you use these classes in 1d and let us know where the
> next
> roadblock is? Maybe we can jointly figure out what to do about things.
>
> Best
> Wolfgang
>
> -------------------------------------------------------------------------
> Wolfgang Bangerth email:
> [email protected]
> www:
> http://www.math.tamu.edu/~bangerth/
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
