From: "Brian Schröder" <ruby.brian@...>
Date: 2005-08-31T04:11:39+09:00
Subject: Re: [ANN] Priority Queue 0.0.0.0. ... .0

On 30/08/05, Levin Alexander <levin@grundeis.net> wrote:
>  [snip]
> > ## Usage
> >
> >     require 'priority_queue'
> >
> >     q = PriorityQueue.new
> >     q.push "node1", 0
> >     q.puts "node2", 1
> 
> Why not "def push(priority, *args)"?  That would make it easier to store
> multiple things in the queue.

What would that mean? Create n nodes with priority priority, or create
on node with an array of n entries. I think that would be unintuitive.

> 
>    q.push distance, x, y
>    [...]
>    x, y = q.pop_min
> 
> >     q.decrease_priority("node2", -1)
> 
> It should be possible to add things to the queue multiple times and access
> them separately.  This would not work in the current version:
> 
>    q.push :foo, 1
>    q.push :foo, 100
>    q.decrease_priority :foo, 0
> 
> If you wrap everything into a new array when it is inserted into the queue
> the following would be possible:
> 
>    h1 = q.push 1, :foo    #=> [:foo]
>    h2 = q.push 100, :foo  #=> [:foo]
>    q.decrease_priority(0, h1)
> 
> I don't know if this is better, though

I could expose the nodes of the priority queue. I decided against
this, because that means that I can spare creating a ruby object for
each queue entry.
Maybe I should create two different types of queues, one with the hash
magick I'm using already and one where the user has to keep track of
the nodes himself.

> 
> 
> Viele Gr��e,
> Levin
> 
> 

Danke,

Brian

-- 
http://ruby.brian-schroeder.de/

Stringed instrument chords: http://chordlist.brian-schroeder.de/