Re: [boost] Input caching iterator adaptor

From: John E. Potter (jpotter_at_[hidden])
Date: 2001-07-06 15:47:48

On Fri, 6 Jul 2001, David Abrahams wrote:

> > Let's ask the same irrelevant question about random access iterator.
> 
> It's not quite the same question, since you're not printing the value_type
> of the iterator. I understand that in your example the notion of equivalent
> sillys is a bit, um, silly though.

OK, I'm convinced.  Input iterator may have a value type of void.  The
operator== returns true.  All algorithms work because all ranges are
empty.  There is absolutely nothing in the standard which requires an
input iterator to be useful.

Does the statement in forward iterator that a == b demands that
*a == *b require forward iterator value_type to be equality comparable?

I think we all know what equivalent should mean in input iterator.  Is
there really a problem in the standard, or is there a problem in trying
to create something which does not iterate over a sequence and claim
that it is an iterator?  What purpose does that claim serve?

John