on Sat Nov 01 2008, Joaquin M Lopez Munoz <joaquin-AT-tid.es> wrote:
Not that this has to do with the multi-indexing capabilities of Boost.MultiIndex, but I think you can still take (a little) advantange of the lib by using a multi_index_container<hashed_unique> instead of a unordered_map, because the former has an iterator_to() member function that allows you to retrieve an iterator from a value_type*, so saving you the jump from Key* to an iterator via map.find() that your code has at pop().
I am using boostpro-1.34.1-svn for this project. Since it is based on boost-1.34.1, it doesn't have iterator_to yet :(. I could backport the multi_index changes, but in reality all I need is a plain unordered_map. If I could prevail upon Daniel James to add iterator_to to Boost.Unordered so it will be there in the future, I'd gladly backport that change into my version -- Dave Abrahams BoostPro Computing http://www.boostpro.com