2008-08-06
| 15:13 | ericthor | rhickey: anything change today related to classpath or resource lookups? |
| 15:35 | slava | rhickey: how does the performance of your persistent hash maps compare to java hash maps, in particular for updates? |
| 16:14 | rhickey | ericthor: no |
| 16:15 | rhickey | slava: about 2-4x slower for inserts, can be as fast or faster for lookups, depending on number of items |
| 16:16 | slava | ah, ok |
| 16:19 | slava | that's pretty much what i got with my implementation, but i'm hoping to close the gap with compiler optimizations |
| 16:19 | slava | but man, persistent hash maps are pretty hairy to implement |
| 16:19 | slava | it took me a while to understand how BitmapIndexedNodes work -- bitCount(bitmap & (bit - 1)) is a bad-ass trick |
| 16:42 | rhickey | slava: sorry about that :) |
| 16:43 | slava | no worries :) |
| 16:44 | slava | so what's next for clojure? |
| 16:45 | rhickey | slava: mostly refinement over the next few months - October will be a year in release, might be nice to call something 1.0 |
| 16:46 | rhickey | longer-term, some sort of integrated Prolog-like rule engine, and a distributed computing story |
| 16:46 | rhickey | I hope for web frameworks etc from contributors |
| 16:47 | slava | rule engine? interesting |
| 16:48 | rhickey | slava: makes more of a program declarative |