dolfin team mailing list archive

["Garth N. Wells" <g.n.wells@xxxxxxxxxx>]

Quoting Anders Logg <logg@xxxxxxxxx>:

> On Fri, Mar 24, 2006 at 11:01:29PM +0100, Garth N. Wells wrote:
> > I've tried some more, and this is where I'm at:
> > 
> > 1. When using the Intel C++ compiler with the gcc header files (which is
> > required in order to link with other libraries compiled with gcc), the
> > only fatal error is the problem with PArray.h. This I haven't worked
> > this out yet. There are a lot of warnings which I can fix. A warning
> > which I can't fix (needs to be done in FFC) relates to the virtual
> > functions  
> >   void BiLinearForm::real [], const AffineMap&, uint) const
> >   void LinearForm::real [], const AffineMap&, uint) const
> > 
> > The error is
> > 
> > error #654: overloaded virtual function "dolfin::BilinearForm::eval" is
> > only partially overridden in class
> > "dolfin::ConvectionDiffusion2D::BilinearForm"
> >   class BilinearForm : public dolfin::BilinearForm
> > 
> > Could we put an empty eval() function which takes 3 arguments in the FFC
> > generated files?
> 
> Is the problem that the two functions have the same name and one is
> overloaded but not the other?
> 

Yes.

> I can fix this in FFC, it's a simple fix to generate the extra
> function.
> 
> > 2. When using the Intel C++ compiler with the Intel header files, I get
> > a lot more errors, which can also be fixed. One thing is when using
> > STLs, the header files need to be included, which the aren't now. For
> > example, in MeshInit.cpp, the function set_intersection() is called,
> > which I couldn't find anywhere in DOLFIN. I realised later the it could
> > be fixed by
> >   #include<algorithm>
> > Why don't we need to include this when using g++?
> > 
> > I think that sorting out this case in particular will enhance
> > portability.
> > 
> > Garth
> 
> I don't recognize the set_intersection() stuff, didn't even know it
> was there (most likely Johan J has added this). But yes, all the STL
> headers should be included where appropriate. I don't know why gcc
> does not complain.

I think that these header files are possibly included via the gcc header files
since I only get the error when using non-gcc header files. I'll add them where
needed.

Garth

> 
> /Anders
> 
> 
> > On Fri, 2006-03-24 at 14:14 -0600, Anders Logg wrote:
> > > On Fri, Mar 24, 2006 at 07:01:44PM +0100, Garth N. Wells wrote:
> > > > Another problem is in DefaultParameters.h. Throughout the code, I've
> had
> > > > to include header files for Standard Template Library. In
> > > > DefaultParameters.h I get the error mesage
> > > > 
> > > > /dolfin/DefaultParameters.h(57): error: namespace "std" has no member
> > > > "numeric_limits"
> > > >   add("homotopy maximum size", std::numeric_limits<int>::max());
> > > >                                     ^
> > > > 
> > > > ./dolfin/DefaultParameters.h(57): error: type name is not allowed
> > > >   add("homotopy maximum size", std::numeric_limits<int>::max());
> > > > 
> > > > but if I add #include<limits> I then get a whole bunch of other
> errors.
> > > > 
> > > > Garth
> > > 
> > > Where did you include <limits>? In ParameterSystem.cpp?
> > > 
> > > /Anders
> > > 
> > > _______________________________________________
> > > DOLFIN-dev mailing list
> > > DOLFIN-dev@xxxxxxxxxx
> > > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> > 
> > 
> > _______________________________________________
> > DOLFIN-dev mailing list
> > DOLFIN-dev@xxxxxxxxxx
> > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> > 
> 
> -- 
> Anders Logg
> Research Assistant Professor
> Toyota Technological Institute at Chicago
> http://www.tti-c.org/logg/
> 
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
>