Re: reference type of output_iterator_wrapper

[Joe Buck <Joe dot Buck at synopsys dot COM>]

On Tue, Apr 18, 2006 at 10:08:15AM -0400, Douglas Gregor wrote:
> 
> On Apr 17, 2006, at 3:59 PM, Howard Hinnant wrote:
> >Fwiw I suggested the standard do something like this in Berlin and  
> >got met with pretty strong opposition (give a non-void  
> >iterator_traits to output iterators).
> 
> I obviously missed that part of the discussion.
> 
> >The standard currently says that value_type and difference_type  
> >should be void for output iterators (24.3.1/1).  I think it's a bad  
> >rule, and apparently I'm not alone.  But there's an up hill battle  
> >in the LWG here.  Packaging it with concepts might help.
> 
> Well, the right thing to do with value_type and difference_type for  
> output iterators is to say nothing at all about them. void is  
> blatantly wrong, because:
> 
> 	- if it's an output iterator, difference_type must be void
> 	- if it's a forward iterator, difference_type must be a signed  
> integral type
> 	- every forward iterator is an output iterator
> 	- but void isn't a signed integral type

I assume, then, that this is bad wording and is meant to apply only to
output iterators that are not also forward iterators.  That is, it's
supposed to be a way to make templates that expect forward iterators
fail cleanly on output iterators.  Otherwise, as you say, there are
conflicting requirements.