|
|
Ublas : |
Subject: Re: [ublas] ublas + move semantics
From: Norman Goldstein (normvcr_at_[hidden])
Date: 2011-12-27 11:42:55
That is an interesting work-around, but I want to write my
code for the (hopefully soon to be) mainstream C++11 syntax.
I am wondering what are the plans to upgrade Boost numeric vectors
and matrices to C++11 in this respect?
Thank you,
Norm
On 12/27/2011 04:40 AM, Nasos Iliopoulos wrote:
> Norman,
>
> for non C++11 copy elision - that has a similar effect as C++11 move
> semantics - you can use the compiler option
> BOOST_UBLAS_MOVE_SEMANTICS (for gcc -DBOOST_UBLAS_MOVE_SEMANTICS).
>
> See the following for more details:
> http://www.boost.org/doc/libs/1_41_0/libs/numeric/ublas/doc/options.htm
>
> Best,
> Nasos
>
> On 12/25/2011 01:28 PM, Norman Goldstein wrote:
>> I am using boost 1.46.0 on fedora 15, using the -std=c++0x
>> flag on gcc, and found that the rvalue reference move semantics
>> are not supported. I put the following code into vector.hpp and
>> matrix.hpp, and valgrind is happy.
>>
>> ======== vector.hpp
>> BOOST_UBLAS_INLINE
>> vector (vector &&v):
>> vector_container<self_type> (),
>> data_ (NULL){ assign_temporary(v); }
>>
>>
>> ========= matrix.hpp
>> BOOST_UBLAS_INLINE
>> matrix ( matrix&& m):
>> matrix_container<self_type> (),
>> size1_ (m.size1_), size2_ (m.size2_), data_ (NULL) {
>> assign_temporary(m); }
>>
>>
>> I am not sure if the above code is correct, never mind optimal.
>> What are the plans for C++0x in boost/numeric?
>>
>> Thank you.
>>
>> PS I first checked the latest boost version, 1.48, and did
>> not see the support, there, either.
>> _______________________________________________
>> ublas mailing list
>> ublas_at_[hidden]
>> http://lists.boost.org/mailman/listinfo.cgi/ublas
>> Sent to: athanasios.iliopoulos.ctr.gr_at_[hidden]
>
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas
> Sent to: normvcr_at_[hidden]
>