Re: [boost] Re: Re: pointer_traits ?

From: Howard Hinnant (hinnant_at_[hidden])
Date: 2002-09-24 18:08:27

On Tuesday, September 24, 2002, at 07:18  PM, Bohdan wrote:

>
> "Douglas Gregor" <gregod_at_[hidden]> wrote in message
> news:200209241302.48886.gregod_at_cs.rpi.edu...
>> On Tuesday 24 September 2002 09:04 am, Peter Dimov wrote:
>>> From: "Bohdan" <warever_at_[hidden]>
>>>
>>>> Hmm ... , looks like my problem is too exotic :)
>>>
>>> No, it's not that exotic, the indirect iterator adaptor has the same
>>> problem. The answer to your original question is that yes, std/boost 
>>> is
>>> lacking pointer_traits (or dereference_traits) (consider the feature
>>> request acknowledged), and C++ is lacking typeof(*p).
>>
>> ... or just iterator_traits extended to allow trivial iterators.
>> boost/iterator_traits.hpp, anyone?
>
> IMHO i like this idea. The simplest solutions are always the best , but
> are you sure that this two idioms (iterators & pointers ) should be 
> mixed ?
> Does standart allow such iterators ?
>
> PS: not too serious:
> If  "anyone" implements iterator_traits for a smart_ptr, than
> what will be the name for its iterator category:
> "uninterative_iterator_category" ?

I keep expecting Jeremy to jump in on this one, but I'll do it for him. 
  Check out Jeremy Siek's Improved Iterator Categories and Requirements:

http://anubis.dkuug.dk/jtc1/sc22/wg21/docs/papers/2001/n1297.html

I think the idea of separating out iterator traversal from dereference 
return types is very interesting.  Then smart pointers could have 
dereference traits but not traversal traits.

-Howard