Re: Cycle Iterator Adapter

From: rogeeff (rogeeff_at_[hidden])
Date: 2002-02-05 16:35:13

--- In boost_at_y..., Thomas Witt <witt_at_i...> wrote:
> > m_bounds.second - m_bounds.first could be negative. But 
cycle_size_()
> > is always returing positive number. What is cycle_size you 
mentioning?
> 
> Is this really true. I think all methods assume second >= first and 
I think 
> thats ok. m_bounds should be required to be a valid half_open range.

Oops. I are right of course. I do not know where did I get an idea 
that they could be in wrong order. I uploaded fixed version.

> 
> I hadn't time to check it but I really feel uncomfortable with the 
advance 
> code. Gennadiy are you aware of the fact that advance, despite its 
name, 
> might be called with negative argument? That one fooled me. I hope 
to give it 
> a closer look tomorrow

Yes, advance will work correctly for both negative and positive n, 
independently on operator% implementation. The only requirement is 
like this:
   For any y and any z > 0: x = y % z => |x| < z

> 
> Thomas

Gennadiy.