Re: [Boost-users] Asymmetrical predicates in STL algorithms

Dave, Thanks! Isn't there a problem with equal_range() where both the first edition of the standard and the work paper you referred to explicitly mention comp(e,value) and comp(value,e)? That, while consistent with the behavior of the original STL, seems to still require a symmetrical predicate... Regards, ...Max... "David Abrahams" wrote in message news:u8xsb4vx7.fsf@boost-consulting.com...

"Max Motovilov" writes:

...

Now, this:

struct less { bool operator()( const Object&, const Key& ) const; bool operator()( const Key&, const Object& ) const; };

lower_bound( v.begin(), v.end(), Key(something), less() );

should work everywhere, or so I think. Would be nice to be able to create such an adaptor for a pair of functions (or a pair of bind or lambda expressions for that matter). I blogged a simple solution (for functions only) here

http://swsw.motovilov.net/weblogs/swsw/archives/2006/02/asymmetrical_pr.html

...

but am wondering if perhaps I missed a generic solution in one of the Boost libraries, or there's one coming in the new release?

Well, standard libraries are now required to make it legal:

http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#270

And you can get an implementation where it works at

http://www.boost-consulting.com/vault/index.php?&direction=0&order=&directory=Algorithms/binary_search

HTH, -- Dave Abrahams Boost Consulting www.boost-consulting.com