From: jhrwalter (walter_at_[hidden])
Date: 2001-12-08 07:30:27
--- In boost_at_y..., Toon Knapen <toon.knapen_at_s...> wrote:
> Peter Schmitteckert (boost) wrote:
>
> > Salut,
> >
> > On Friday 07 December 2001 13:37, Toon Knapen wrote:
> >
> >>Peter Schmitteckert (boost) wrote:
> >>
> >>>Allthough a submatrix adapter would be sufficient for me.
> >>>
> >>Do you mean that you would first create a matrix of sufficient
size,
> >>then take a submatrix and resizing the latter does not need
reallocation
> >>as long as the size of the submatrix is not bigger as the size of
the
> >>original matrix. That would indeed cover the `quest for a minimal
number
> >>of reallocations`.
> >>
> >
> > Yes, but I think a submatrix should not be allowed to resized
larger than
> > the original matrix.
>
>
> Agreed. But what should happen if the users tries it anyway : throw
an
> exception ? But isn't testing for this too time-consuming as the
> intention is exactly to have quickly a matrix with the correct size
> without losing time for allocating a new one.
Our general strategy in ublas is, to test such conditions in debug
mode and to disable these checks in release mode.
Regards
Joerg