I observe the following in Python 2.6.1 in CentOS 5.9 and in Python 2.7.5 in MacOS 10.9.2.

A heapified 3-element array containing elements in the order [low, high, medium] fails to print in sorted order, as shown below: 

>>> import heapq
>>> lista = [1, 6, 5, 4]
>>> heapq.heapify(lista)
>>> lista # Gives sorted order, as expected
[1, 4, 5, 6]
>>> 
>>> listb = [1, 6, 5]
>>> heapq.heapify(listb)
>>> listb # Gives unsorted order -- unexpected and presumably a bug
[1, 6, 5]
>>> 
>>> [heapq.heappop(listb) for i in range(len(listb))] # But heappop works
[1, 5, 6]

Hi,
I dont think its a bug. 
The textbook definition of a min(or max) heap is

"Heaps are binary trees for which every parent node has a value less than or equal to any of its children."

Therefore,
when lista = [1,6,5,4], and heapify is run on it,it can be viewed as
    1
  /   \
4  5  6

or [1,4,5,6]

When listb=[1,6,5], running heapify on it,gives
   1
  / \
 6   5

or [1,6,5]
heapify maintains the heap-invariant - Every key is smaller than its children. This invariant is not violated in the above example.

The heappop maintains the heap-invariant after popping. The line in your code
[heapq.heappop(listb) for i in range(len(listb))]
is the heapsort algorithm!.

Hopefully, this clears your question.

Hi sahutd,

You may be right. I was keying off a stackoverflow example (http://stackoverflow.com/questions/12373837/heapq-module-python) that might have been misleading.

In fact, the Python documentation never does say what a printed heapified array is suppose to look like -- how it unwinds the tree structure into a linear list.

But from the above example I assumed it was supposed to print the sorted order, and again my assumption is quite possibly wrong.

It might be useful, as a documentation issue, to say what a printed heapified array is supposed to look like.

It's not really relevant what a heapified list looks like (and there is no reason to guarantee a particular appearance, since it's an implementation detail that could change). It's supposed to function as a heap with the heap functions, that's all. The docs do give a guarantee that a sorted list is already "heapified", but that's a one way guarantee: All sorted lists are heaps, but not all heaps are sorted lists.

The docs also mention the big-O behavior of heapify; it's linear time, O(n), while good general purpose sorting algorithms are O(n log n). A linear algorithm cannot sort a general list.

By the way, in the top answer to the Stack Overflow post you linked, it's clear that the list wasn't sorted. [44, 42, 3, 89, 10] became [3, 10, 44, 89, 42], and you'll notice, the last three elements are out of order.

All fair enough. I see the error of my ways. I am closing the case with a Resolution of "invalid." Thanks....