On Fri, Mar 03, 2006 at 11:34:58AM +0300, Bulat Ziganshin wrote:
Friday, March 3, 2006, 1:10:17 AM, you wrote:
JM> Would it be possible to mock up an implementation of this using JM> Data.Generics by chance? so we can use a generalized tree for anything
are you know that Data.Generics has run-time overheads and therefore can be several times slower than fixed algorithm or algorithm generated at compile time (for example, by TH) ?
JM> <noise> JM> once get get derived generalized tries, I want to try derived JM> generalized huffman (arithmetic?) encoding, derived generalized LZ*, and JM> a derived generalized efficient 'diff' algorithm... JM> </noise>
and this will be 1000 times slower that C implementations
Oh yes. I am completly aware of that. I just find the algorithms fascinating. DrIFT would be the obvious real way to go for haskell (or TH if you are into that sort of thing) or perhaps c-generated haskell... -- John Meacham - ⑆repetae.net⑆john⑈