Re: [Boost-users] [Boost.MPI] question about serializingcustomobject

Subject: Re: [Boost-users] [Boost.MPI] question about serializingcustomobject
From: Matthias Troyer (troyer_at_[hidden])
Date: 2010-08-27 15:55:36

On 27 Aug 2010, at 21:25, David Abrahams wrote:

> At Fri, 27 Aug 2010 21:05:38 +0200,
> Matthias Troyer wrote:
>> 
>> 
>> On 27 Aug 2010, at 19:44, Robert Ramey wrote:
>> 
>>> David Abrahams wrote:
>>>> At Thu, 26 Aug 2010 16:05:41 -0800,
>>>> Robert Ramey wrote:
>>>>> I just never found it to be very revealing.  The word skeleton
>>>>> seemed pretty suggestive. It's still not clear to me how such a
>>>>> think can work between heterogeneous machines.  For example, if I
>>>>> have an array of 2 byte integers and they each need to get
>>>>> transformed one by one into a 4 byte integer because that's closest
>>>>> MPI data type,
>>> 
>>>> I think you don't understand what MPI datatypes do.
>>> 
>>> This is true.  I suppose that's one reason why the documentation
>>> made no sense to me. They just looked like special types
>>> to make identify primitives accross differing architectures.
>>> 
>>> <snip>
>>> Interesting information which one should consider adding to the
>>> MPI documentation.
>>> </snip>
>> 
>> I disagree. We cannot and should not copy dozens of pages of MPI
>> documentation. Boost.MPI is targeted at an audience that understands
>> MPI but does not presume to teach MPI.
> 
> Dozens of pages?  Copy?  Teach MPI?
> 
> AFAICT, Robert is only suggesting adding 83 lines (~4K characters) to
> the docs.  And frankly, even though I understand and have used MPI,
> none of what I wrote seems obvious to me, and frankly it seems
> important in explaining to people why they should use Boost.MPI rather
> than trying to use MPI directly.

The question is where should we stop, where should we start, what should be explained. The MPI standard documents is several hundred pages long. If you want just a few lines discussing the advantages of Boost.MPI and of skeleton&content then this can be done. If you want us to explain all details of MPI datatypes, all details of the semantics of nonblocking operations (which caused other issues in the past), ....then it will soon be most of the MPI standard or an MPI text book.
> 
> Frankly, I'd like to put that information in a blog post, if I got it
> right (did I?), so you wouldn't have to do more than point to it, even
> though IMO it's ill-advised not to explain some of this right in the
> MPI doc.

The docs were written by Doug but he's probably too busy at the moment.

Matthias