Hello Matias,
thank you for the solution.
1. Is the bimap method as efficient as using, for example,
the hashmap form the gnu-ext namespace ? ( I am talking about time-efficiency)
2. Is a bimap serializeable by boost::serialize ? That would be absolutely
essential
3. I am sure that I read somewhere boost multi-index can be used as a mormal hashmap
Did I misinterpret something ? i am asking because I know that multiindex is serializeable.

regards, Oliver

On 8/21/07, Matias Capeletto <matias.capeletto@gmail.com> wrote:
On 8/21/07, Oliver Kania <kania.oliver@googlemail.com> wrote:
> Hello,

Hi Oliver,

> what I basically want to do is using multi-index like a hashmap.
> I want to use vector<long> as key. However, all the examples I found
> are using member-fields of the values or the values themselves as key.
> Is it not possible to use multi-index like a hash_map<value,key> where the
> key is independent of the value type ?
> If yes, could someone please provide an example ?

You need Boost.Bimap. Online docs here: http://tinyurl.com/22sja5
Look at "Unconstrained sets" section.
It will be included in Boost 1.35, you can check out of svn head now.

An example of what you are asking:

typedef
   bimap
   <
      vector_of< long >,
      unconstrained_set_of< something >
   >
   bm_type;

bm_type bm;
assign::push_back( bm ) ( 1, a ) ( 2, b ) ( 3, c );

BOOST_FOREACH( bm_type::left_reference p, bm.left )
{
    cout << p.first << "-->" << p.second;
}

King regards
Matias
_______________________________________________
Boost-users mailing list
Boost-users@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users