Re: [Boost-Users] Re: Efficient implementation of associative_property_map

From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-03-05 16:39:22

At 08:14 AM 3/5/2003, Tarjei Knapstad wrote:

 >On Wed, 2003-03-05 at 13:50, okrslar wrote:
 >> --- In Boost-Users_at_[hidden], Jeff Holle <jeff.holle_at_v...> wrote:
 >> > I'd think a hashed dictionary would be needed for such a large data
 >set.
 >> > There is a caviot though, weither or not a good hash function can be
 >> > created.
 >> >
 >> > The good thing is that its easy to experiment with different STL
 >> > associative containers, as long as you have SGI's STL or a 
derivative/
 >> > or good alternative.
 >> >
 >> > In SGI, the container is hash_map.
 >>
 >> We don't have SGI STL around here. I've seen that before, but I hoped
 >> that there would be an other way than changing my STL version... It's
 >> amazing that there is no hashtable in the stdc++, isn't it?
 >>
 >More or less... :) Hashed containers were introduced too late to make it
 >into the standard IIRC, allthough most STL implementations have them and
 >most STL books document them.
 >
 >I believe you can be pretty certain they'll make it into the next
 >standard though :)

Yes, Matt Austern has written a nice hash table proposal for the upcoming 
Standard Library Technical Report. He plans to present version 3 of his 
proposal in Oxford in April, and I'd be very surprised if it doesn't get 
accepted, and probably at this meeting.

--Beman