From: Stefano Crocco <stefano.crocco@...>
Date: 2007-09-18T19:08:27+09:00
Subject: Re: how to use tuple as hash key

Alle marted狸 18 settembre 2007, SpringFlowers AutumnMoon ha scritto:
> SpringFlowers AutumnMoon wrote:
> > but my question is, if i run the loop 10 million times (for some
> > statistics), then it will actually duplicate the array 10 million times,
> > which is memory hungry.
>
> actually, i tried running it and watch ruby.exe in Windows or ruby in
> Mac keep on being 2 or 3MB without every growing.   (i set the loop to
> 100 million and dup the key and watched it keep on running with the
> program size about the same while running).
>
> I think it is due to the hash table's size is constant.  so when Hash
> called hash() to find the slot, and do the eql? and find that they are
> the same.  and so it either places that dup key in that slot or don't
> place it there at all (doesn't matter), and the un-referenced
> dupulicates are garbage collected.  so that's why the size of the
> program keep on being 2 to 3MB in the RAM.
>
> if however, i set an array  just_test  and  always do a  just_test <<
> key   where key is the dup, then the program will grow from 3MB to 6MB,
> to 12MB, to 20MB and it just keeps growing. (just_test is an array to
> keep a reference to the dup objects so that they won't be garbage
> collected).   So looks like the garbage collection in Ruby is doing a
> good job.

I think it's easier (and more efficient) to create a new array with the 
current values of i and j at each iteration, instead of modifying one single 
array and duping it.

for i in 1..2
  for j in 1..2
    h2[[i,j]] = "ha"
  end
end

Stefano