in reply to Re: Data Structure needed for Event Queue
in thread Data Structure needed for Event Queue

Like a number of others in the monastery, I would suggest you not use Heap. This may be one of those rare cases where it's faster and more reliable to write and debug your own version than go to CPAN. Alternatively, the authors of "Elements of Programming with Perl" have been kind enough to put a kinder, gentler implementation on the web. I haven't tested it, but it's at least simple enough that it's possible to read and understand the code, and relatively painless to use it.

/s

  • Comment on Re: Re: Data Structure needed for Event Queue

Replies are listed 'Best First'.
Re: Re: Re: Data Structure needed for Event Queue
by demerphq (Chancellor) on May 28, 2002 at 15:03 UTC
    Just curious, but without my references im not really sure but what kind of heap is that? Fibonnacci?

    Yves / DeMerphq
    ---
    Writing a good benchmark isnt as easy as it might look.

[reply]
      Just curious, but without my references im not really sure but what kind of heap is that? Fibonnacci?

      Assuming you are referring to the EoPwP heap that educated_foo points to, then (as stated by educated_foo, who refers to me in the plural :-) it is strictly a simple, array-based implementation of a binary heap. It isn't documented except as it is built in the book.

      A main difference between this and the Heap model is that this one contains the comparison routines internally and works on the model of selecting a comparison method on heap creation and supplying a suitable key and satellite data for each inserted element (it is easy to use any objects for which you can supply a key) --- to use objects with the Heap version you need all of your objects to supply their own cmp() method for comparing themselves with any other object type in the heap. The latter is more flexible because the comparison routines may be arbitrarily complex, but does require that all your objects either inherit or supply their own common comparison method.

[reply]
      Which "that"? Heap implements normal ("binary"), binomial, and fibonnacci heaps. The "Elements of" version is a normal binary heap.

      /s

[reply]
        Hmm, I was reviewing AoP yesterday and it seemed to suggest that binary heaps arent particularly optimal. Wonder why they picked that one?

        Yves / DeMerphq
        ---
        Writing a good benchmark isnt as easy as it might look.

[reply]

In Section Seekers of Perl Wisdom