Re: [boost] status of fibonnaci_heap and mutable_queue

From: Doug Gregor (dgregor_at_[hidden])
Date: 2006-04-10 15:36:08

On Apr 6, 2006, at 5:10 PM, Fernando Cacciola wrote:
> All I need is a priority queue with an update() operation (which of  
> course
> can potentially replace the top element), so yes, if I understood the
> interface correctly, this is exactly what I need!!
>
> It's also under pending, but I guess this one is actually used in  
> the BGL so
> is trustworthy, right?

We'd done a lot of validation of shortest paths algorithms using the  
relaxed heap as the priority queue, so it should be trustworthy.

> BTW: a related question that it think you can answer:
>
> This DS needs an index map from the IndexedType to an integral  
> index in the
> range [0,n].
>
> But my algorithm will run over non-indexed types which might not  
> even be
> stored in a radom access container.
>
> If I understand the propery maps library correctly, I'll need to  
> create a
> map, or hash_map, associating each and every possible value with an  
> int, and
> wrap that into an associative_property_map<> to input into the  
> relaxed_heap
> DS.
>
> Is that right?

That sounds right.

        Doug