ffc team mailing list archive

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

Anders Logg wrote:
> Didn't know that. Yes, it looks like -f blas is broken for boundary
> integrals. Do you know if it has ever worked?
>

I don't think that it has worked since I added boundary integrals to
FFC. I didn't touch the BLAS output format. It was a while ago, so I
forgot that it wasn't working.

Garth


> /Anders
> 
> 
> On Wed, Sep 06, 2006 at 02:48:16PM +0200, Garth N. Wells wrote:
>> Something else - are thingd working with the -blas option? This wasn't
>> working for me last week. The problem is with the new boundary integrals.
>>
>> Garth
>>
>> Anders Logg wrote:
>>> The interface for the forms has changed slightly in both FFC and
>>> DOLFIN, but if you install the latest version of both (FFC 0.3.3 and
>>> DOLFIN 0.6.2), there should be no problems.
>>>
>>> Nothing in the interface changed from yesterday's snapshot of FFC to
>>> version 0.3.3.
>>>
>>> /Anders
>>>
>>>
>>> On Wed, Sep 06, 2006 at 02:17:38PM +0200, Dag Lindbo wrote:
>>>> Something fishy is going on... I get errors in all .h files generated by
>>>> FFC when I build my DOLFIN-based solver.
>>>>
>>>> I've been working on other things the last week, so I can't be sure that
>>>> it's the new FFC that's causig this. But I did intall the snapshot from
>>>> yesterday evening.
>>>>
>>>> In file included from main.cpp:39:
>>>> forms/LSM_adv_SUPG.h: In constructor
>>>> `dolfin::LSM_adv_SUPG::BilinearForm::BilinearForm(dolfin::Function&,
>>>> dolfin::Function&, dolfin::Function&, const dolfin::real&)':
>>>> forms/LSM_adv_SUPG.h:506: error: cannot allocate an object of type
>>>> `dolfin::LSM_adv_SUPG::BilinearForm::TestElement'
>>>> forms/LSM_adv_SUPG.h:506: error:   because the following virtual
>>>> functions are abstract:
>>>> /afs/nada.kth.se/home/c/u1d4g64c/PDC/temp07/local/include/dolfin/FiniteElement.h:56: error:  virtual void dolfin::FiniteElement::vertexeval(dolfin::real*, unsigned int, const dolfin::real*, const dolfin::Mesh&) const
>>>> forms/LSM_adv_SUPG.h:509: error: cannot allocate an object of type
>>>> `dolfin::LSM_adv_SUPG::BilinearForm::TrialElement'
>>>> forms/LSM_adv_SUPG.h:509: error:   because the following virtual
>>>> functions are abstract:
>>>> /afs/nada.kth.se/home/c/u1d4g64c/PDC/temp07/local/include/dolfin/FiniteElement.h:56: error:  virtual void dolfin::FiniteElement::vertexeval(dolfin::real*, unsigned int, const dolfin::real*, const dolfin::Mesh&) const
>>>> forms/LSM_adv_SUPG.h:512: error: cannot allocate an object of type
>>>> `dolfin::LSM_adv_SUPG::BilinearForm::FunctionElement_0'
>>>> forms/LSM_adv_SUPG.h:512: error:   because the following virtual
>>>> functions are abstract:
>>>> /afs/nada.kth.se/home/c/u1d4g64c/PDC/temp07/local/include/dolfin/FiniteElement.h:56: error:  virtual void dolfin::FiniteElement::vertexeval(dolfin::real*, unsigned int, const dolfin::real*, const dolfin::Mesh&) const
>>>>
>>>> regards,
>>>> Dag Lindbo
>>>>
>>>> _______________________________________________
>>>> FFC-dev mailing list
>>>> FFC-dev@xxxxxxxxxx
>>>> http://www.fenics.org/mailman/listinfo/ffc-dev
>>> _______________________________________________
>>> FFC-dev mailing list
>>> FFC-dev@xxxxxxxxxx
>>> http://www.fenics.org/mailman/listinfo/ffc-dev
>>>
>> _______________________________________________
>> FFC-dev mailing list
>> FFC-dev@xxxxxxxxxx
>> http://www.fenics.org/mailman/listinfo/ffc-dev
> _______________________________________________
> FFC-dev mailing list
> FFC-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/ffc-dev
>