On 8/4/07, Elmer Fittery <[EMAIL PROTECTED]> wrote: > It isn't so much the fastest sort that I was interested in, but rather > the "trie" structure itself. The sort is fast because of the "trie" > structure.
As I understand it, a lot of this speed issue is about the application -- its access patterns and the granularity of the objects it's working with. In other words, I thinik there's some assumptions here that the operations are "word at a time" instead of "a whole bunch of words at a time". Even there, the big point seems to be that "lists longer than length 2 seem to be more efficient, in some applications involving sorting, than lists of length 2". Anyways... because this seems to be application oriented I'm dubious that it belongs in the language. (Which I think I remember someone else suggesting.) -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
