Re: [Boost-mpi] Thread API

Subject: Re: [Boost-mpi] Thread API
From: Matthias Troyer (troyer_at_[hidden])
Date: 2012-12-31 05:01:06

Hi Alain,

This basically replicates the functionality of the C API. I think I would still like to see support for automatic throwing of exceptions if the required level could not be obtained. I see two variants: either a wrapper boost::mpi::require(boost::mpi::funneled), or further additional levels, like require_funneled. I have a tendency towards the former since it means that the levels are still the same as that of the C API and we clearly identify the requirement. Does anyone have other suggestions?

Matthias

On Dec 4, 2012, at 10:40 AM, Alain O Miniussi <alain.miniussi_at_oca.eu> wrote:

> To sum up what I got from the comments regarding the threading API, do
> we agree that we only want to basically to mimic the underlying C-Init,
> that is:
> - provide an environment ctor with only one preferred threading support
> level.
> - provide a query method to probe the provided support level (user code
> will need to check if it is ok with it).
> - scan the documentation and make sure that any usage of such a ctor is
> followed by a checking of the provided level ?
> - if no level is provided, assume "threading::single" (which we do not
> need to test, since that the minimal level)
> 
> If that ok, I'll try to provide a patch accordingly.
> 
> Thanks
> 
> 
> On Thu, 2012-11-29 at 12:04 -0600, Hal Finkel wrote:
>> ----- Original Message -----
>>> From: "Matthias Troyer" <troyer_at_[hidden]>
>>> To: "alain miniussi" <alain.miniussi_at_oca.eu>
>>> Cc: "Discussion of Boost.MPI development" <boost-mpi_at_[hidden]>
>>> Sent: Thursday, November 29, 2012 7:21:09 AM
>>> Subject: Re: [Boost-mpi] Thread API
>>> 
>>> 
>>> On Nov 29, 2012, at 1:11 PM, Alain O Miniussi <alain.miniussi_at_oca.eu>
>>> wrote:
>>>> 
>>>> That being said, I do not have a strong opinion between:
>>>> a) provide a 2 argument ctor
>>>> b) provide one argument ctor, and test provided level afterward.
>>>> 
>>>> I just have a preference for b, because I find it simpler (both in
>>>> terms
>>>> of API and explanation, one thing I did not mention is that we
>>>> could as
>>>> weel abort if the requirement is not met in the ctor, since we
>>>> won't be
>>>> able to do any MPI related stuff after that, so going into a lot of
>>>> trouble to explain the status of env seems unjustified), and that
>>>> throwing in ctor has it's own set of issue (if only being
>>>> controversial).
>>> 
>>> Actually throwing in the constructor if we cannot acquire a resource
>>> (here MPI with required level of thread support) is just the RAII
>>> idiom. There are no problems with that and it is actually good
>>> style.
>>>> 
>>>> One thing to consider is that if the requirement is not met, we
>>>> probably
>>>> want to explain what's going on to the user. Which will require
>>>> some
>>>> code. So the test "if (provided < required) { ... }" won't add a
>>>> lot of
>>>> trouble anyway.
>>> 
>>> I agree. If the requested level is not met we will need to test
>>> anyway which level we got.
>> 
>> Fair enough. Maybe the best thing to do is just to make sure that we actually do this test is all of the documentation snippets and example code.
>> 
>> -Hal
>> 
>>> 
>>> Matthias
>>> 
>>> 
>>> _______________________________________________
>>> Boost-mpi mailing list
>>> Boost-mpi_at_[hidden]
>>> http://lists.boost.org/mailman/listinfo.cgi/boost-mpi
>>> 
>> 
> 
> -- 
> Alain Miniussi
> Responsable Tech. Centre de Calcul Haute Performance
> Obs. de la Côte d'Azur |Mont Gros: [phone]
> BP 4229                |Sophia :   [phone]
> 06304 Nice Cedex 4     |https://crimson.oca.eu
> 
> _______________________________________________
> Boost-mpi mailing list
> Boost-mpi_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-mpi