Question

I am currently in the thinking process of the data structure I might use for a current project. I don’t need to delete items as i am loading a database,using it, and then exiting the program. The only constraint concerns search time.(memory in a second time but mainly time).

Overview on what I intend to do.
I am parsing files and extracting informations that I use to create various object.
After reading the files and creating the objects,i have a set of multiple objects that have references to other one as string.

The goal here is to find which net goes from one domain to another

E.g : text input file :

module Blabla 
netTomodule Foo
domain 1
..../*Other parameters of the module*/
end module

module Foo 
netTomodule Blabla
netTomodule Foo2
domain 2
..../*Other parameters of the module*/
end module

module Foo2
netTomodule Foo
domain 2
..../*Other parameters of the module*/
end module

After reading this I get 3 module objects Foo Foo2 and Blabla and their attributes are as follow :

class Module{
private :
string name;
int domain;
netlist * mynetlist;
...
}  

My opinion and the thing i want to get advice on :

After thinking about this, i think that my best shot is to :

1. When reading the file and extracting info, i should create a linked list of Module.
2. Then with the number of Module i have read, i create an array that is double the size of that.
3. For each module, I use a hash function to hash the module name and put a pointer to this module at the given index in the array
4. Now when i will want to find a module, i just have to compute the hash value and get the pointer at the given index(or increment if its not the good module because of a collision previously in the making of the array)

This is basically an implematation of a hashtable or atleast what i know of a hashtable from my clasess.

My question is Is this a good thought ? Is there a hashtable library i can use that does that ? (i have heard and look for unordered_map and map but i don’t know if it fits my needs very well)

This is a huge text so i hope it is detailed enough, so thank you if you have the courage to read everything !

Answer 1

Just use any hash table that comes with your standard library or from boost. Most will have the unordered_map (as specified by TR1 and proposed for C++0x) as does boost, but some will have a std::hash_map or stdext::hash_map with various implementation being slightly different, e.g. the original SGI vs. Microsoft.

You don’t need to build a list, just put the objects directly in the hash table; it allows sequential iteration, though it will be in some fixed random order.