Re: [Haskell] [Haskell-cafe] ANN: unordered-containers - a new, faster hashing-based containers library

[Johan Tibell]

On Sat, Feb 19, 2011 at 4:49 PM, wren ng thornton  wrote:
> How does, or will, this package differ from Milan Straka's
> http://hackage.haskell.org/package/hashmap ?

It's a continuation of that work. The unordered-containers
implementation is more space efficient and a bit faster. Milan also
uses a Patricia trie in his implementation. I plan to switch to a
hash-array mapped trie (HAMT) in the future. It's about 25% faster for
lookups but at the moment quite a bit slower for inserts. Once we got
some new faster array copy primops in GHC we can hopefully improve the
insert time a lot. A HAMT also has much lower space overhead per
entry.

Johan

___
Haskell mailing list
Haskell@haskell.org
http://www.haskell.org/mailman/listinfo/haskell

Re: [Haskell] [Haskell-cafe] ANN: unordered-containers - a new, faster hashing-based containers library

[Jason Dusek]

  What are your thoughts on iterative construction of maplike
  datastructures? Could something like builder work for maps,
  too? In the project I'm working on, I have a function that
  receives a bunch of YAML fragments and builds a big YAML map
  out of them.

--
Jason Dusek
Linux User #510144 | http://counter.li.org/

___
Haskell mailing list
Haskell@haskell.org
http://www.haskell.org/mailman/listinfo/haskell