std::map perfomance

std::map is a good solution.

Kuphryn

I expect std::map<> is fast enough for your needs. There is also the (non-standard, iirc, but I believe it's in STLport) std::hash_map<>, which should be quite easy to plug in, if you find later that std::map<> isn't fast enough for your needs.

I would strongly suggest that you continue to use std::map<> until you actually have emperical evidence that it is too slow for your purposes (meaning, you're using your scripting language for something, it isn't fast enough, and you've done a bit of profiling to make sure that it's the std::map<> in your scripting language that's the bottleneck). If you do eventually find that std::map<> isn't fast enough, then try replacing it with std::hash_map<> or your own hash table code.

Edit: lol. In the time it takes me to write a reply like this, two people can give one-line answers containing the bulk of what I'm trying to say.

John B

[edited by - JohnBSmall on April 18, 2004 7:08:25 PM]

std::map is generally implemented as a red-black tree, so it is O(log n) for access, insertion and deletion. It is very useful if you need your elements in a sorted order, but if you don't a better fit would be std::hash_map, which is a hash table implementation of map (ie. amortized O(1) access, insertion and deletion times).

If you haven't seen big-O notation around a lot, O(log n) means that the operation cost is proportional to the log of the amount of data you have stored in the structure. O(1) means it executes in constant time, so the number of elements is mostly irrelevant.

However, std::hash_map is sometimes not included in older versions of the STL, if it isn't take a look at STLPort (hit google).

[Edit: okay, too slow on the draw ]

[edited by - premandrake on April 18, 2004 7:08:36 PM]

thanks to everyone for quick and quality answers, thats all i wanted to know, thanks..

bio