#haskell - Mon 19 Feb 2007 between 22:23 and 22:39

NY Lost Funds



soreardons: 50% through nobench ...
donsCale, yeah. needs some tweaks
chessguydons, have a moment to answer a question about the map fusion article?
donsmaybe :)
chessguy(we're talking about http://www.randomhacks.net/articles/2007/02/10/map-fusion-and-haskell-performance for those who haven't seen it)
dons?where hbc
lambdabothttp://www.cs.chalmers.se/~augustss/hbc.html
dons?where+ hbc http://www.cs.chalmers.se/~augustss/hbc/hbc.html
actionsorear poisons his nobench results with elinks startup CPU
sorearpoisons his nobench results with elinks startup CPU
chessguywhy can't he just write a new function treeMap' f g = treeMap (f . g)
lambdabotDone.
donschessguy: he could. but then what about fusing folds/ filters/ ... ?
and he'd then have to manually find all occurences of treeMap f . treeMap g
even is users of the libraries...
chessguyah, i see
donsalso, ghc might move two maps together, even if they're not together in the src
which will now fuse, given a rewrite rule
chessguyinteresting
donsso the rewrite rules tells ghc how to optimise specific uses of the library api
meaning a simpler api, you don't have to provide treeMap' for example
chessguygotcha
thanks
CaleProviding treeMap' is kind of silly anyway, as it only saves about 1 character of typing to use it.
chessguyhmm, i bet such optimizations would do a lot for chess engines
by the way, Cale unfoldTreeM worked like a charm in my program-tree generation code
sorear14/24
chessguy> "Get a calculator"
lambdabot"Get a calculator"
actionddarius still has to read dons' paper, but then he has about 60 others in his queue.
ddariusstill has to read dons' paper, but then he has about 60 others in his queue.
actionsorear won't read dons' paper until nobench finishes, but will concurrent implement time(1) support. Emacs, strangely enough, uses far less CPU than X.
sorearwon't read dons' paper until nobench finishes, but will concurrent implement time(1) support. Emacs, strangely enough, uses far less CPU than X.
donssor yeah, time support sounds good.
Cale> map (chr . fromIntegral) . unfoldr (\x -> guard (x > 0) >> return (x `mod` 256, x `div` 256)) $ 152110696948989857750358645756541560135
lambdabot"Get a calculator"
LoganCapaldo_LOL
soreardons: sor\t eh?
funny, time(1) is missing a PORTABILITY section
donssorry. benchmark made things kind of slow on my laptop. :)
sorearnice!
:)
mwc_Is there a clean way to use read and handle bad input?

Page: 5 12 19 26 33 40 47 54 61 68 75 82 89 

IrcArchive

NY Lost Funds