sorry, Dons, ---------- Forwarded message ---------- From: Alberto G. Corona Date: 2008/11/18 Subject: Re: [Haskell-cafe] implementing python-style dictionary in Haskell To: Don Stewart By the way byteStrings are wonderful, but, it isn´t true that byteStrings are not so fast for managing short strings, for example keys ? 2008/11/18 Don Stewart dave:

...

2008/11/18 kenny lu :

...

Here is a comparison of memory usage

Map : 345 MB IntMap : 146 MB Trie : 282 MB Python : 94 MB

Here is a comparison of execution time (on an intel dual core 2.0G)

Map: 26 sec IntMap: 9 sec Trie: 12 sec Python: 2.24 sec

The above number shows that my implementations of python style dictionary are space/time in-efficient as compared to python.

Can some one point out what's wrong with my implementations?

This isn't really a fair comparison. Map, IntMap, and Trie are persistent data structures, and Python dictionaries are ephemeral. (That is, when you "add" a key to a Map, you actually create a new one that shares structure with the old one, and both can be used in subsequent code. In Python, you would have to copy the dictionary.)

Strings, not ByteStrings. that's the difference.

-- Don _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe