Steven Schveighoffer wrote:
On Tue, 01 Jun 2010 09:17:15 -0400, Don <nos...@nospam.com> wrote:
Steven Schveighoffer wrote:
On Sun, 30 May 2010 18:33:22 -0400, Andrei Alexandrescu
<seewebsiteforem...@erdani.org> wrote:
The heap is a tad difficult to tackle. Most of the time you don't
want to create a heap, but instead to organize an existing range as
a heap. As such, the heap is not always obvious to think of as a
container. I'm undecided on how to approach this.
It's easier to think of a heap as a single entity with operations on
it. At least for me anyway.
Most of the time, once you make a range a heap, you want to continue
to use it as a heap. Restricting operations on that range by
defining a heap type around it can do this. Otherwise, you could
accidentally do something foolish like sort the range.
-Steve
But for several graph algorithms, (eg, A* pathfinding), you have
{key1, key2} pairs, forming a heap based on key1, but you also need to
able to search for key2.
The container is a hybrid, consisting of heap on {key1} + AA on {key2}.
It uses the heap operations, but it's not exactly a heap.
And such a container would be not a heap. I don't think it should be
impossible to define such a construct if a vanilla heap type is also
defined.
My recommendation: keep the heap functions in std.algorithm and create a
std.container based off them.
That's what I think, too.
