Re: lifetime of ranges vs. iterators

On Fri, Aug 29, 2008 at 1:50 PM, Arno Schödl <aschoedl <at> think-cell.com>wrote:

> For iterators, the make_ helpers are great:
>
> vector<int> vecnA;
> vector<int> vecnB;
>
> DoSomethingWithRangeThatStoresIterators(
>        make_difference_iterator( vecnA.begin(), vecnA,end(), vecnB.begin(),
> vecnB.end() ),
>        make_difference_iterator( vecnA.end(), vecnA,end(), vecnB.end(),
> vecnB.end() )
> );
>
> The equivalent with ranges, supposedly easier, would now look like this:
>
> vector<int> vecnA;
> vector<int> vecnB;
>
> difference_range< int, int > diffrng( vecnA, vecnB ); // watch out with
> scope
> DoSomethingWithRangeThatStoresIterators( diffrng );
>
> as opposed to the much nicer:
>
> vector<int> vecnA;
> vector<int> vecnB;
>
> DoSomethingWithRangeThatStoresIterators( make_difference_iterator( vecnA,
> vecnB ) );
>
> If we drive this stacking to one or more levels, things pretty quickly
> become pretty ugly. Do we really want that?
>

No, we don't want that at all. Apologies for the shameless plug, but you
might want to look at the range adaptors that use the '|' operator in the
Boost.RangeEx proposal / update in the vault at:
http://www.boostpro.com/vault/index.php?action=downloadfile&filename=range_ex.zip&directory=Algorithms

The range adaptors and the pipe operator allow infix composition of range
adaptors for the very reason you suggest.

Neil Groves

>
> Arno
>
>
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost