Den Fri, 02 Mar 2007 20:33:45 +0100 skrev Diez B. Roggisch: > Thomas Dybdahl Ahle schrieb: >> I have a sort function in a python chess program. Currently it looks >> like this: >> >> def sortMoves (board, table, ply, moves): >> f = lambda move: getMoveValue (board, table, ply, move) >> moves.sort(key=f, reverse=True) >> return moves >> >> However I'd really like not to use the lambda, as it slows down the >> code. >> >> I've thought about saving the extra variables in the global space, but >> it really feals ugly. >> >> Do you have any ideas how I can sort these moves the fastest? > > First of all, in your case it is somewhat strange to use > f = lambda ... > because then you could as well use > def f(move): > ....
Wouldn't that be just as slow? > But that is just a general remark. Regarding the question: I don't see > how that could possibly become faster without much more insight into > what you are doing in getMoveValue. As it seems, it is dependend of a > lot of factors that change often, so caching it isn't a real option. And > I hope you are aware that the key-method is invoked only _once_ per > list-item! Yeah, key is a nice thing. My only problem is that I need these other objects to generate the value, and I don't want to create a new function each time.. In my profiling the functions with the lambda line says 860 cumtime and getMoveValue says 580. -- http://mail.python.org/mailman/listinfo/python-list
