[Dune-devel] On which side is DataHandleIF::size() called

[Martin Nolte]

Hi Andreas,

there was a discussion and we had a proposal. I think I have some photographs of 
the black board, but I will have to look. Basically, the idea was to have a new, 
more flexible interface for the data handle (and deprecate the old one).

Best,

Martin

On 07/26/2017 09:34 AM, Dedner, Andreas wrote:
> I have a great idea:
> A small group should get together to discuss the parallel interface.
> Cheers
> Andreas
> 
> 
> 
> Sent from my Samsung Galaxy smartphone.
> 
> -------- Original message --------
> From: Martin Nolte <nolte at mathematik.uni-freiburg.de>
> Date: 26/07/2017 08:28 (GMT+00:00)
> To: dune-devel at lists.dune-project.org
> Subject: Re: [Dune-devel] On which side is DataHandleIF::size() called
> 
> Hi Markus,
> 
> you assumption seems correct to me: In general, the receiver cannot know how
> much data was sent. Loadbalancing for a discrete function space with
> element-dependent polynomial degree would be a simple example.
> 
> Unfortunately, there are many situations in which the receiver does actually
> know how much data is sent (e.g., Finite Volume, i.e., 1 DoF per element), but
> there is no way to communicate this to the grid.
> 
> There are further deficiencies of the current interface in more complex
> scenarios (e.g., how do you sent data of different type?) that cannot currently
> be handled. Two years ago, we started discussing over a new concept for a data
> handle. Maybe it is finally time to make it happen.
> 
> Best,
> 
> Martin
> 
> On 07/26/2017 09:18 AM, Markus Blatt wrote:
>> Hi,
>> 
>> when I first had to deal with parallelizing a DUNE grid I somehow
>> assumed that the size method is only allowed to be called on the
>> sending side. Why else what the scatter method have the size as an
>> additional parameter.
>> 
>> Somehow I was not sure whether that assumption really holds or every
>> grid developers thinks so. I just wrote a handle to be used for
>> loadbalancing a grid from a master process to all. As a precaution I
>> introduced an assertion in the size method that makes sure it is only
>> called on rank 0. For UGGrid this assertion just failed.
>> 
>> Is my assumption wrong, is it a bug in UG or am I missing something
>> fundamental here?
>> 
>> Cheers,
>> 
>> Markus
>> 
> 
> -- 
> Dr. Martin Nolte <nolte at mathematik.uni-freiburg.de>
> 
> Universität Freiburg                                   phone: +49-761-203-5630
> Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
> Hermann-Herder-Straße 10
> 79104 Freiburg, Germany
> 
> _______________________________________________
> Dune-devel mailing list
> Dune-devel at lists.dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-devel

-- 
Dr. Martin Nolte <nolte at mathematik.uni-freiburg.de>

Universität Freiburg                                   phone: +49-761-203-5630
Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
Hermann-Herder-Straße 10
79104 Freiburg, Germany