Re: [ublas] revised patch to control initialization of vector

From: Neal Becker (ndbecker2_at_[hidden])
Date: 2005-08-19 07:46:43

Michael Stevens wrote:

> On Wednesday 17 August 2005 23:41, Neal Becker wrote:
>> Here is a revised patch.  I think I prefer this method to control
>> initialization (pass an argument to vector constructor)
> 
> In your patch you have
> 
>          ~unbounded_array () {
>              if (size_) {
> -           std::_Destroy (begin(), end(), alloc_);
> -           alloc_.deallocate (data_, size_);
> +                const iterator i_end = end();
> +                for (iterator i = begin (); i != i_end; ++i) {
> +                    iterator_destroy (i);
> +                }
> +                alloc_.deallocate (data_, size_);
>              }
>          }
> 
> Any ideas where _Destroy comes from and what it does. I assume it is not
> part of the STL.

I have done a quick benchmark on this.  It appears that in at least one
simple case I tried, it didn't make any difference to use this patch.  I
suppose gcc-4.0.1 is smart enough to eliminate the destruction of Complex
values without the optimization of using std::_Destroy.  Therefore, I
withdraw this part of the patch.