On Nov 3, 1:51 pm, [EMAIL PROTECTED] wrote: > Arnaud Delobelle: > > > Here's another idea, probably more practical: > > >>> sorted(s, key=lambda x: tuple(256-ord(l) for l in x), reverse=True) > > Nice. > A variant that probably works with unicode strings too: > > print sorted(s, key=lambda x: [-ord(l) for l in x], reverse=True) > > Bye, > bearophile
Awesome! I tested it on a sample list of ~61K words [1] and it's almost 40% faster, from ~1.05s dropped to ~0.62s. That's still >15 times slower than the default sorting (0.04s) but I guess there's not much more room for improvement. George [1] http://www.cs.pitt.edu/~kirk/cs1501/Pruhs/Spring2006/assignments/boggle/5desk.txt -- http://mail.python.org/mailman/listinfo/python-list
