A case for valueless AA's?

[spir]

On 03/30/2011 05:02 AM, dsimcha wrote:
> On 3/29/2011 8:37 PM, Jonathan M Davis wrote:
>> We now have std.container.RedBlackTree, which can be used as a set, but it is
>> a tree rather than a hash table.
>>
>> - Jonathan M Davis
>
> This isn't a very good set. It doesn't even support basic set primitives like
> intersection, difference, union, etc. Furthermore, most programs will probably
> want a hash set. A tree set is only better when you don't have a good hash
> function or worst case performance is more important than average case.
>
> I'm not saying there's anything wrong with RedBlackTree. It's a perfectly good
> data structure to implement a set on top of. It's just that it's not a "real"
> set type because it's not meant to be.

Yep, but with balanced binary search trees you get sorting for free ;-)

Denis
-- 
_________________
vita es estrany
spir.wikidot.com