[Dune] Fwd: Re: Communicating IdSet elements

[Martin Nolte]

Hi Jö,

while we might learn from Boost how to do seriaization properly, I don't think a
dependency on boost a good idea. Furthermore, glancing over the
Boost::Serialization documentation, they are intended to be non-intrusive, i.e.,
given "good" operations for conversion to and from binary, you could use boost
on top of dune-grid. I would very much prefer such an approach over dependencies
on boost.

Best,

Martin

On 07/24/2011 07:32 PM, Jö Fahlke wrote:
> Am Sun, 24. Jul 2011, 17:09:51 +0200 schrieb Martin Nolte:
>> -------- Original Message --------
>> Subject: Re: [Dune] Communicating IdSet elements
>> Date: Sun, 24 Jul 2011 17:09:29 +0200
>> From: Martin Nolte <nolte at mathematik.uni-freiburg.de>
>> To: Jö Fahlke <jorrit at jorrit.de>
>>
>> Hi Jö,
>>
>> with respect to disk I/O the lack of precision was indented. I guess having a
>> good binary representation is all you need to achieve both, communication and
>> any kind of disk I/O (you can always write in in base64 or hexadecimal
>> encoding into a character stream). But I wanted to keep an open discussion at
>> the current state.
>>
>> As to the communication: Maybe Markus can give further insights on the current
>> status, here. Moreover, I think there might be some good ideas in the MPI
>> standard that can also help for arbitrary serialization problems (read: disk
>> I/O above).
> 
> I didn't read the standard recently, but as far as I can remember, you always
> need to register new data types such as structs with mpi so it knows how to
> deal with them.  AFAIK Boost.MPI solves the issue by using Boost.Serialization
> (or whatever that library is called exactly), that is why I mentioned
> serialization previously.  With Boost.Serialization the issue boils down to
> making the IdType serializable, i.e. adding a sort of dump() and load()
> functionality.  The down side is that this would add a dependency on Boost.
> On the other hand we probably don't want to roll our own serialization system.
> 
> Boost.Serialization was first part of Boost 1.32.0 which was released November
> 19th, 2004.  I don't know whether any significant incompatible changes
> happened in the meantime.
> 
> Are there any other candidates for serialization systems we could use?
> 
> Bye,
> Jö.
> 

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

Universität Freiburg                                   phone: +49-761-203-5642
Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
Hermann-Herder-Straße 10
79104 Freiburg, Germany