Re: [boost] py_cpp and vector<..>::const_iterator

From: Ullrich Koethe (koethe_at_[hidden])
Date: 2000-12-01 06:10:35

David,

yes, it would be more natural, but also less efficient if the original
C++ sequence didn't support random access. Suppose, l was
std::list<Foo>. Then 

     for x in l.cursor():

takes constant time per step (linear overall), while

     for x in l:

takes linear time per step (quadratic overall). This difference occurs
because cursor remembers the last item retrieved (by way of its embedded
iterator), while list doesn't. This only reinforces Stepanov's great
insight that working with iterators leeds to more generic code than
working with data structures.

For C++ containers that support indexing (including std::map, hash_map),
we could provide automatic wrapping into the more natural interface.
However, I don't see an urgent enough need for this feature to implement
it now.

Regards
Ulli

David Abrahams wrote:
> 
> Ullrich,
> 
> the cursor stuff looks good, but it leaves me wondering why your interface
> provides:
> 
>     y = l.cursor()[3]
>     for x in l.cursor():
> 
> but not:
> 
>     y = l[3]
>     for x in l:
> 
> which would seem more natural.
> 

-- 
 ________________________________________________________________
|                                                                |
| Ullrich Koethe  Universität Hamburg / University of Hamburg    |
|                 FB Informatik / Dept. of Computer Science      |
|                 AB Kognitive Systeme / Cognitive Systems Group |
|                                                                |
| Phone: [phone]                Vogt-Koelln-Str. 30 |
| Fax:   [phone]                D - 22527 Hamburg   |
| Email: u.koethe_at_[hidden]               Germany             |
|        koethe_at_[hidden]                        |
| WWW:   http://kogs-www.informatik.uni-hamburg.de/~koethe/      |
|________________________________________________________________|