[Patch] First bits of the algo merge

[Peter Dimov]

Howard Hinnant wrote:
> On Dec 14, 2005, at 10:51 AM, Howard Hinnant wrote:
>
>>  That proxy test is still on the way...
>
> Ok, below is a test.  ...

> struct equal_to
> {
>     template <class _Lhs, class _Rhs>
>     bool operator()(const _Lhs& __lhs, const _Lhs& __rhs) const

Need _Rhs here.

> {return __lhs == __rhs;}
> };

>     public:
>         operator T& () const {return iter_->dirty_deref();}
>         operator const T& () {return iter_->deref();}

I don't get it. A non-const proxy is convertible to T const& and a const 
proxy converts to T&?