On Behalf Of Corwin Joy
I think we should support both an input iterator and a random access iterator.
I think I understand why there are problems with any iterator more powerful than an input iterator and I agree. I don't understand what benefit there is to supporting anything besides input iterators though. I've always used databases that provided random access, for instance, but I can't think of anything I've ever done with a database that couldn't have been done with just input iterators. To keep things simple I'd suggest just supporting input iterators. If we do support random access iterators, then the database library must either be templatized or have two sets of begin/end functions, one of which may fail at runtime. I think making it a template would be better.