Do you mean this?
namespace boost {
template< ... > It lower_bound( It begin, It end, T value ) { using boost::partition_point; return partition_point( begin, end, boost::bind( ... ) ); }
template< ... > It partition_point( It begin, It end, Pred pred ) { return std::lower_bound( begin, end, boost::bind( ... ) ); }
// a custom implementation of partition_point: template< ... > transform_iterator<It,Func> partition_point(transform_iterator<It,Func> begin, transform_iterator<It,Func> end, Pred pred ) { ... }
Something like that, yes.
See http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#270 Also, see boost/detail/binary_search.hpp, which was introduced precisely to work around one implementation's strict concept checking.
Ok.
Why do you think ADL is o.k. for customizing boost::partition_point (soon std::partition_point, N2666), but not for customizing std::lower_bound? AFAIK, the same arguments apply to both.
Because nobody wants to customize lower_bound, upper_bound, equal_range, and binary_search when they could instead customize partition point alone.
The counting_iterator example above shows that people may want to, for special cases and not inside a library, but as an optimization for their special iterator implementation. It feels funny to me to use different techniques for different functions. And you pay the overhead of boost::bind for any call to lower_bound, customized or not. I am willing to do it, though, if that's what you prefer. Arno -- Dr. Arno Schoedl · aschoedl@think-cell.com Technical Director think-cell Software GmbH · Invalidenstr. 34 · 10115 Berlin, Germany http://www.think-cell.com · phone +49-30-666473-10 · toll-free (US) +1-800-891-8091 Directors: Dr. Markus Hannebauer, Dr. Arno Schoedl · Amtsgericht Charlottenburg, HRB 85229