Re: [boost] reorder_iterator : iterator_adaptor

From: Toon Knapen (toon.knapen_at_[hidden])
Date: 2001-11-27 08:17:25

David Abrahams wrote:

>>OK, the notation is not optimal ;-). The idea is just that ultimatly the
>>user would only need to create a past-the-end permutation iterator with
>>one argument, the past-the-end of the order. This is simpler for the
>>user and less confusing, since otherwise the user needs to construct the
>>past-the-end from two iterator of which one is actually not used (and
>>thus what value should it have ideally ?)
>>
> 
> OK. Downside: all end-of-sequence iterators using the same permutation but a
> different element set look identical, thwarting bug detection somewhat.

That's true. Also, the constructor of iterator_adaptors would need to be
specialised for the case where the policies class is a 
permutation_iterator_policies to make sure the first arguments goes into 
the policies and the second argument is used as the base. This also gets 
ugly.