Re: [eigen] STL compatible aligned_allocator + StdVector

[Gael Guennebaud <gael.guennebaud@xxxxxxxxx>]

On Thu, Apr 9, 2009 at 8:20 PM, Hauke Heibel
<hauke.heibel@xxxxxxxxxxxxxx> wrote:
> Hi again,
>
> It seems that the aligned_allocator is missing two important operators - see
> also here for further details
> http://blogs.msdn.com/vcblog/archive/2008/08/28/the-mallocator.aspx.
>
> bool operator!=(const aligned_allocator<T>& other) const
> {
>   return !(*this == other);
> }
>
> // Returns true if and only if storage allocated from *this
> // can be deallocated from other, and vice versa.
> // Always returns true for stateless allocators.
> bool operator==(const aligned_allocator<T>& other) const
> {
>   return true;
> }

thanks, patch applied.

> I think that at least the second one is required (by std::swap) in order to
> get the code from StdVector compiled under VS 2008 (vc9). Furthermore, the
> ei_workaround_msvc_std_vector might be extended by
>
> template <typename OtherT>
> inline ei_workaround_msvc_std_vector(const Eigen::MatrixBase<OtherT>& other)
> : T(other) {}
>
> This should allow the initialization via expressions like (taken from
> BVH.cpp)
>
> v.push_back(VectorType::Random());
>
> This still does not solve all problems for non MatrixBase derived types
> (e.g. AlignedBox) - simply because the constructor interface will not be
> inherited by ei_workaround_msvc_std_vector. Maybe template class
> specializations can help out.

I'm not sure to understand what's going one,  the compilation errors would help.

> I won't be answering to any responses of this mail since I am on holidays!
> Wish you all a happy Easter weekend.

happy Easter to you too

> Regards,
> Hauke
>