On Thu, 23 Apr 2009 19:09:14 -0400, Mike Meyer wrote:

If you believe this is still true (though as has been pointed out, this idea has already been considered and rejected), show us the timings. In particular, with a range of different values for the percentage of excluded elements: say 0, 10, 50, 90 and 100?

Ok, I'll admit when I'm wrong. It actually takes longer to use the list comprehension (for all exclusion percentages), which, for me at least, is completely unintuitive. For 1 million samples: excluded: 0%, comprehension time: 0.840000 excluded: 0%, if time: 0.530000 excluded: 10%, comprehension time: 0.770000 excluded: 10%, if time: 0.520000 excluded: 20%, comprehension time: 0.710000 excluded: 20%, if time: 0.490000 excluded: 30%, comprehension time: 0.670000 excluded: 30%, if time: 0.460000 excluded: 40%, comprehension time: 0.610000 excluded: 40%, if time: 0.450000 excluded: 50%, comprehension time: 0.560000 excluded: 50%, if time: 0.420000 excluded: 60%, comprehension time: 0.510000 excluded: 60%, if time: 0.400000 excluded: 70%, comprehension time: 0.450000 excluded: 70%, if time: 0.380000 excluded: 80%, comprehension time: 0.410000 excluded: 80%, if time: 0.350000 excluded: 90%, comprehension time: 0.360000 excluded: 90%, if time: 0.330000 excluded: 100%, comprehension time: 0.310000 excluded: 100%, if time: 0.300000 I hereby accept the rejection of this idea since it makes sense neither from a syntax nor efficiency point of view. Best Regards, Mike