Re: [ublas] Incomplete initialization in hermitian_matrix_element

From: Georg Baum (Georg.Baum_at_[hidden])
Date: 2006-07-26 04:48:15

Am Mittwoch, 26. Juli 2006 10:41 schrieb Gunter Winkler:
> this is the patch
>
> Index: hermitian.hpp
> ===================================================================
> RCS file: /cvsroot/boost/boost/boost/numeric/ublas/hermitian.hpp,v
> retrieving revision 1.56
> diff -u -p -r1.56 hermitian.hpp
> --- hermitian.hpp	6 Nov 2005 14:57:13 -0000	1.56
> +++ hermitian.hpp	26 Jul 2006 08:39:56 -0000
> @@ -60,7 +60,7 @@ namespace boost { namespace numeric { na
>              container_reference<matrix_type> (m), i_ (i), j_ (j), d_
> (d), dirty_ (false) {} BOOST_UBLAS_INLINE
>          hermitian_matrix_element (const hermitian_matrix_element &p):
> -            container_reference<matrix_type> (p), i_ (p.i_), d_ (p.d_),
> dirty_ (p.dirty_) {} +            container_reference<matrix_type> (p),
> i_ (p.i_), j_ (p.j_), d_ (p.d_), dirty_ (p.dirty_) {} BOOST_UBLAS_INLINE
>          ~hermitian_matrix_element () {
>              if (dirty_)

May I ask why this copy constructor exists at all? With your patch it is 
equivalent to the code the compiler would generate if no copy constructor 
existed. That seems safer to me, because the compiler does not forget to 
initialize members.
Apart from that, what about

hermitian_matrix_element &operator = (const hermitian_matrix_element &p)?

Does it need to copy i_ and j_ as well?

Georg