[Dune] [Dune-Commit] dune-grid r4163 - trunk/grid/common

[Oliver Sander]

Hi!
I fell into the same trap as Markus for 10 seconds.
On first reading it looks like

return this->boundary();

and that would indeed segfault.  But with the extra
dereference

return (*this)->boundary();

it doesn't.

:-)
Oliver

Andreas Dedner schrieb:
> I do not understand why the method should segfault.
> Do you have some code where this happens? All tests
> work as well as our dune-fem stuff which is not yet
> ported to the Intersection class.
> (*this)-> gives the Intersection class where the Methods
> have to be implemented and it should not be the
> same boundary() method as on the IntersectionIterator
> because in that case one would get a DEPRECATED
> warning even when calling the boundary method correct
> on the Intersection class.
> IIter iter;
> iter->boundary() 
> has to work without DEPRECATED warning and nothing else
> (as I see it) happens in Martins fix. He is implementing
> the functionality using the new implementation...
>
> It would be nice if you could give us a piece of code that
> does not work because all our examples even with
> iter.boundary()
> all work fine.
> Best Andreas
>
> On Fri, 23 May 2008 07:46:46 +0200
>  Markus Blatt <Markus.Blatt at ipvs.uni-stuttgart.de> wrote:
>   
>> On Thu, May 22, 2008 at 10:45:22PM +0200, Martin Nolte
>> wrote:
>>     
>>> It is like this. (*this) is a reference to
>>>       
>> IntersectionIterator< ... >. Then, 
>>     
>>> *(*this) is a reference to the Intersection< ... >.
>>>       
>> Then, on the intersection, 
>>     
>>> the boundary method calls the implementation (at least,
>>>       
>> this is the way I 
>>     
>>> understand it).
>>>
>>>       
>> And you intended this?
>>
>> Of what use is a deprecated method that segfaults if
>> called? It is
>> just annoying.
>>
>> In my opinion a deprecated method should work and
>> definitely not
>> segfault! 
>>
>> I would recommend, either undoing the changes or start a
>> discusion
>> whether to remove the deprecated methods in question.
>>
>> Or if possible provide the deprecated functionality by
>> calling
>> undeprecated methods.
>>
>> Cheers,
>> Markus
>>
>> _______________________________________________
>> Dune mailing list
>> Dune at dune-project.org
>> http://lists.dune-project.org/mailman/listinfo/dune
>>     
>
>
> _______________________________________________
> Dune mailing list
> Dune at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune
>   


-- 
************************************************************************
* Oliver Sander                ** email: sander at mi.fu-berlin.de        *
* Freie Universität Berlin     ** phone: + 49 (30) 838 75217           *
* Institut für Mathematik II   ** URL  : page.mi.fu-berlin.de/~sander  *
* Arnimallee 6                 ** -------------------------------------*
* 14195 Berlin, Germany        ** Member of MATHEON (www.matheon.de)   *
************************************************************************