Re: [ublas] Vector of size 3 without the size element

From: Nico Galoppo (nico_at_[hidden])
Date: 2007-05-29 04:39:12

Can't you just subclass c_vector and provide an accessor to data_ ? Some pseudo 
code:

template <typename T>
class vector3 : public ublas::c_vector<real_type, 3> {
..
public:
   array_type::value_type* data() { return data_; }
..
}

Then you can do

typedef vector3<float> vector3f;
vector3f p(0.0f, 1.0f, 1.0f);
glVector3fv(p.data());

or, I think you should be able to do the following with a regular c_vector

c_vector<float, 3> p(0.0f, 1.0f, 1.0f);
glVector3fv(&p(0));

HTH,

--nico

Langeland, Stian (GE Healthcare) wrote:
> Thanks for such a quick reply. I have considered the c_vector but when I
> looked in vector.hpp I see that it has a member size_
> ....
>     private:
>         size_type size_;
>         array_type data_;
> .... 
> I guess this will still come in-between the floats?
> 
> Stian
> 
>> -----Original Message-----
>> From: ublas-bounces_at_[hidden] 
>> [mailto:ublas-bounces_at_[hidden]] On Behalf Of Dima Sorkin
>> Sent: 29. mai 2007 10:11
>> To: ublas mailing list
>> Subject: Re: [ublas] Vector of size 3 without the size element
>>
>> Hi.
>>  It can be c_vector, from
>> http://www.boost.org/libs/numeric/ublas/doc/types_overview.htm
>>
>> I don't sure that if you put an array of c_vector, then all
>> floating numbers will sit continuously in memory (compiler 
>> can add some
>> spaces).
>>
>> May be a 3xN matrix fits your needs ?
>>
>> Regards,
>>  Dima.
>>
>> Quoting "Langeland, Stian (GE Healthcare)" <Stian.Langeland_at_[hidden]>:
>>
>>> Hi,
>>>
>>> I have been trying to create a vector of 3 elements (a 
>> point) without
>>> the size element stored in the object. The reason for this is that I
>>> need to be able to have all the elements consecutive in memory when
>>> gathering several points in a std::vector. In this way all 
>> points in the
>>> std::vector can be uploaded to a graphics card using a 
>> single memcopy.
>>> Can this be acheived using ublas? I have not been able to 
>> find a way.
>>> Thanks,
>>> Stian
>>> _______________________________________________
>>> ublas mailing list
>>> ublas_at_[hidden]
>>> http://lists.boost.org/mailman/listinfo.cgi/ublas
>>>
>> _______________________________________________
>> ublas mailing list
>> ublas_at_[hidden]
>> http://lists.boost.org/mailman/listinfo.cgi/ublas
>>
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas

-- 
Nico Galoppo        UNC-CH PhD. student        http://www.ngaloppo.org
                       [phone]